Data wrangling, grouping and aggregation

Data wrangling, grouping and aggregation#

Next, we will continue working with weather data but expand our analysis to cover longer periods of data from Finland. In this section, you will learn various useful techniques in pandas to manipulate, group, and aggregate the data in different ways that are useful when extracting information from your data. At the end of this section, you will learn how to create an automated data analysis workflow that can be repeated with multiple input files that have a similar structure. As a case study, we will investigate the claim that the summer of 2021 was exceptionally warm in Finland [1].

Cleaning data while reading#

In this section we are using weather observation data from Finland that was downloaded from NOAA (check the data overview section for more details). The input data are separated by varying number of spaces (i.e., fixed column widths). The first lines of the data look like following:

STATION           ELEVATION  LATITUDE   LONGITUDE  DATE     PRCP     TMAX...
----------------- ---------- ---------- ---------- -------- -------- ----...
GHCND:FIE00142226         24    60.2028    24.9642 20051213 -9999    40  ...
GHCND:FIE00142226         24    60.2028    24.9642 20051214 -9999    35  ...
GHCND:FIE00142226         24    60.2028    24.9642 20051215 -9999    38  ...

By looking at the file contents above, we can see a few things that we need to consider when reading the data:

  1. The delimiter: The columns are separated with a varying amount of spaces which requires using some special tricks when reading the data with the pandas .read_csv() function.

  2. The line of dashes: The second line of the data file contains characters separating the column headings from the data.

  3. NoData values: NaN values in the data file are coded as -9999 and hence we need to be able to instruct pandas to interpret those values as NaN.

  4. Unnecessary columns: The input data contains eight columns, and several of those do not contain data we need. Thus, we should probably ignore the unnecessary columns when reading in the data file.

Handling and cleaning heterogeneous input data (such as in our example here) can be done after reading in the data. However, in many cases it is preferable to do some cleaning and preprocessing when reading in the data. In fact, it is often much easier to do things this way. Let’s see how we can handle each point above when reading in the data file.

  1. For our data file, we can read the data with varying number of spaces between the columns by using the parameter sep=r"\s+". In this case, we use a raw text string with the sep parameter, which is indicated by the r before the first quotation mark and ensures the escape character \ is handled properly in the sep string.

  2. The second line of the data file can be skipped using the skiprows parameter, as we have seen earlier. However, this time we will give a list of rows to skip (by index value) so that the header line is read, the second line is skipped, and the rest of the file is read. In this case, we can use skiprows=[1].

  3. For handling the NoData values (point 2 above), we can tell pandas to consider -9999 as NaN by using the na_values parameter and specifying the character string -9999 should be converted to NaN. For this data file we can specify na_values=["-9999"], which will then convert the -9999 values into NaN values.

  4. Finally, we can limit the number of columns that we read (point 3 above) by using the usecols parameter, which we have already used previously. In our case, we are interested in columns that might be somehow useful to our analysis, including the station ID, date, and data about temperatures: 'STATION', 'DATE', 'TMAX', 'TMIN'. Achieving all these things is pretty straightforward using the .read_csv() function, as demonstrated below.

import pandas as pd

# Define relative path to the file
fp = "data/helsinki-kumpula.txt"

# Read data using varying amount of spaces as separator,
# specifying '-9999' characters as NoData values,
# and selecting only specific columns from the data
data = pd.read_csv(
    fp,
    sep=r"\s+",
    skiprows=[1],
    na_values=["-9999"],
    usecols=["STATION", "DATE", "TMAX", "TMIN"],
)

Let’s now check the data by printing the first five rows with the .head() function.

data.head()
STATION DATE TMAX TMIN
0 GHCND:FIE00142226 20051213 40.0 32.0
1 GHCND:FIE00142226 20051214 35.0 26.0
2 GHCND:FIE00142226 20051215 38.0 33.0
3 GHCND:FIE00142226 20051216 36.0 19.0
4 GHCND:FIE00142226 20051217 19.0 15.0

Perfect, looks good. We have excluded some unnecessary columns, skipped the row of dashes, and converted the -9999 values to NaN values (we will confirm this later in this section).

Renaming columns#

Let’s take a closer look at the column names of our DataFrame.

print(data.columns)

Index(['STATION', 'DATE', 'TMAX', 'TMIN'], dtype='object')

As we see, some of the column names could be unclear with only their names. Luckily, it is easy to change labels in a pandas DataFrame using the .rename() function. In order to change the column names, we need to tell pandas how we want to rename the columns using a dictionary that converts the old names to new ones. A dictionary is a specific data structure in Python for storing key-value pairs. We can define the new column names using a dictionary where we list key: value pairs in following manner:

  • STATION: STATION_ID

  • TMAX: TMAX_F

  • TMIN: TMIN_F

Hence, the original column name (e.g., STATION) is the dictionary key which will be converted to a new column name STATION_ID (which is the value). In addition, the temperature values in this data file are represented in degrees Fahrenheit. We will soon convert those temperatures to degrees Celsius. Thus, in order to avoid confusion with the columns, let’s rename the columns TMAX to TMAX_F, and TMIN to TMIN_F. Below we can create a dictionary for the new column names.

new_names = {
    "STATION": "STATION_ID",
    "TMAX": "TMAX_F",
    "TMIN": "TMIN_F",
}
new_names

{'STATION': 'STATION_ID', 'TMAX': 'TMAX_F', 'TMIN': 'TMIN_F'}

Our dictionary format looks correct, so now we can change the column names by passing that dictionary with the parameter columns in the .rename() function.

data = data.rename(columns=new_names)
data.columns

Index(['STATION_ID', 'DATE', 'TMAX_F', 'TMIN_F'], dtype='object')

Perfect, now our column names are easier to understand and use.

Using functions with pandas#

Now it’s time to convert those temperatures from degrees Fahrenheit to degrees Celsius. We have done this many times before, but this time we will learn how to apply our own functions to data in a pandas DataFrame. We will define a function for the temperature conversion and then apply this function for each Fahrenheit value on each row of the DataFrame. The output Celsius values will be stored in a new column called TMAX_C, for example. But first, it is a good idea to check some basic properties of our input data before proceeding with data analysis.

# First rows
data.head(2)
STATION_ID DATE TMAX_F TMIN_F
0 GHCND:FIE00142226 20051213 40.0 32.0
1 GHCND:FIE00142226 20051214 35.0 26.0 
# Last rows
data.tail(2)
STATION_ID DATE TMAX_F TMIN_F
6819 GHCND:FIE00142226 20240830 77.0 61.0
6820 GHCND:FIE00142226 20240831 75.0 57.0 
# Data types
data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 6821 entries, 0 to 6820
Data columns (total 4 columns):
 #   Column      Non-Null Count  Dtype  
---  ------      --------------  -----  
 0   STATION_ID  6821 non-null   object 
 1   DATE        6821 non-null   int64  
 2   TMAX_F      6812 non-null   float64
 3   TMIN_F      6812 non-null   float64
dtypes: float64(2), int64(1), object(1)
memory usage: 213.3+ KB

Nothing suspicious in the first and last rows, but here the .info() function indicates that the number of observations per column varies if you compare the Non-Null Count information to the number of entries in the data (n = 6821). Only the station number and date seem to have data for each row. The other columns appear to have a few missing values. This is not necessarily anything dangerous, but good to keep in mind. Let’s now check the descriptive statistics.

# Descriptive stats
data.describe()
DATE TMAX_F TMIN_F
count 6.821000e+03 6812.000000 6812.000000
mean 2.014896e+07 50.451556 38.707281
std 5.400847e+04 17.302753 15.267066
min 2.005121e+07 -7.000000 -15.000000
25% 2.010083e+07 37.000000 29.000000
50% 2.015050e+07 49.000000 38.000000
75% 2.019123e+07 65.250000 51.000000
max 2.024083e+07 92.000000 74.000000

Looking at the TMAX_F values (maximum daily temperatures in Fahrenheit), we can confirm that our measurements seems more or less valid because the value range of the temperatures makes sense (i.e., there are no outliers such as extremely high or low values) and there are not any -9999 values that were not converted to NaN. It is always a good practice to critically inspect your data before doing any analysis, as it is possible that your data may include incorrect values (e.g., due to a sensor malfunction or human error).

Defining a function#

Now that we are sure that our data looks correct, and we can start our temperature conversion process by first defining our function to convert from Fahrenheit to Celsius. pandas can use regular functions, so you can simply define functions for pandas exactly the same way as you would do normally (as we learned in Chapter 2, for instance). Let’s define a function that converts from degrees Fahrenheit to Celsius.

def fahr_to_celsius(temp_fahrenheit):
    """Function to convert Fahrenheit temperature into Celsius.

    Parameters
    ----------

    temp_fahrenheit: int | float
        Input temperature in Fahrenheit (should be a number)

    Returns
    -------

    Temperature in Celsius (float)
    """

    # Convert the Fahrenheit into Celsius
    converted_temp = (temp_fahrenheit - 32) / 1.8

    return converted_temp

Now we have the function defined and stored in memory. At this point it is good to test the function with a known value to make sure it works properly.

fahr_to_celsius(32)

0.0

32 degrees Fahrenheit is indeed 0 degrees Celsius, so our function seem to be working correctly.

Using a function by iterating over rows#

Next we will learn how to use our function with data stored in a pandas DataFrame. We will first apply the function row by row using a for loop and then we will learn a more efficient way of applying the function to all rows at once.

Looping over rows in a DataFrame can be done in a few different ways. A common approach is to use the .iterrows() method which loops over the rows as index-Series pairs. In other words, we can use the .iterrows() method together with a for loop to repeat a process for each row in a pandas DataFrame. Please note that iterating over rows this way is a rather inefficient approach, but it is still useful to understand the logic behind how this works. When using the .iterrows() method it is important to understand that .iterrows() accesses not only the values of one row, but also the index of the row. Let’s start with a simple example for loop that goes through each row in our DataFrame.

# Iterate over the rows
for idx, row in data.iterrows():
    # Print the index value
    print(f"Index: {idx}")

    # Print the temperature from the row
    print(f"TMIN F: {row['TMIN_F']}\n")

    break

Index: 0
TMIN F: 32.0

We can see that the idx variable indeed contains the index value 0 (the first row) and the row variable contains all the data from that row stored as a pandas Series. Also, notice that when developing a for loop you do not always need to iterate through the entire loop if you just want to test things out. Using the break statement in Python terminates a loop wherever it is placed inside the loop. In our case we used it to check out the values on the first row of the DataFrame. This saves time and allows us to test the code logic without printing thousands of values to the screen!

Next, let’s create an empty column TMIN_C for the Celsius temperatures and update the values in that column using the fahr_to_celsius() function that we defined earlier. For updating the value in the DataFrame, we can use the .at[] indexer that we used earlier in this chapter. This time, however, we will use the .itertuples() method to access the rows in the DataFrame. The .itertuples() method works similarly to .iterrows(), except it returns only the row values without the index. In addition, the returned values are not a pandas Series, but instead .itertuples() returns a named tuple data type. As a result, when using .itertuples() accessing the row values needs to be done a bit differently. Remember, a tuple is like a list but immutable and a “named tuple” is a special kind of tuple object that adds the ability to access the values by name instead of position index. Thus, we can access the TMIN_F value in a given row using row.TMIN_F (in contrast to how we accessed the value in the example above). We will not work with named tuples in the rest of the book, but more information can be found in the Python documentation for named tuples [2].

Let’s see an example of how to use the .itertuples() method.

# Create an empty column for the output values
data["TMIN_C"] = 0.0

# Iterate over the rows
for row in data.itertuples():
    # Convert the Fahrenheit to Celsius
    # Notice how we access the row value
    celsius = fahr_to_celsius(row.TMIN_F)

    # Update the value for 'Celsius' column with the converted value
    # Notice how we can access the Index value
    data.at[row.Index, "TMIN_C"] = celsius

# Check the result
data.head()
STATION_ID DATE TMAX_F TMIN_F TMIN_C
0 GHCND:FIE00142226 20051213 40.0 32.0 0.000000
1 GHCND:FIE00142226 20051214 35.0 26.0 -3.333333
2 GHCND:FIE00142226 20051215 38.0 33.0 0.555556
3 GHCND:FIE00142226 20051216 36.0 19.0 -7.222222
4 GHCND:FIE00142226 20051217 19.0 15.0 -9.444444 
# What does our row look like?
row._asdict()

{'Index': 6820,
 'STATION_ID': 'GHCND:FIE00142226',
 'DATE': 20240831,
 'TMAX_F': 75.0,
 'TMIN_F': 57.0,
 'TMIN_C': 0.0}

Okay, now we have iterated over our data, converted the temperatures to degrees Celsius using our fahr_to_celsius() function, and stored the results in the TMIN_C column. The values look correct as 32 degrees Fahrenheit is 0 degrees Celsius, which can be seen on the first row. We also have the last row of our DataFrame stored in the variable row from the code above, which is a named tuple that has been converted to the dictionary data type using the ._asdict() method for named tuples.

Before moving to other more efficient ways to use functions with a pandas DataFrame, we should note a few things about the approaches above. We demonstrated use of the .itertuples() method for looping over the values because it is significantly faster than .iterrows() (can be around 100 times faster). We also used .at[] to assign the value in the DataFrame because it is designed to access single values more efficiently than the .loc[] indexer, which can access groups of rows and columns. That said, you could have also simply used data.loc[idx, new_column] = celsius to achieve the same result as both examples above. It is simply slower.

Using a function with the apply method#

Although using a for loop with .itertuples() can be fairly efficient, the pandas DataFrame and Series data structures have a dedicated method called .apply() for applying functions in columns (or rows). .apply() is typically faster than .itertuples(), especially if you have a large number of rows. When using .apply(), we pass the function that we want to use as an argument. Let’s start by applying our fahr_to_celsius() function to the TMIN_F column with the temperature values in Fahrenheit.

data["TMIN_F"].apply(fahr_to_celsius)

0        0.000000
1       -3.333333
2        0.555556
3       -7.222222
4       -9.444444
          ...    
6816    15.555556
6817    13.888889
6818    14.444444
6819    16.111111
6820    13.888889
Name: TMIN_F, Length: 6821, dtype: float64

The results again look logical. Notice how we passed the fahr_to_celsius() function without using the parentheses () after the name of the function. When using .apply(), you should always leave out the parentheses from the function that you use. In other words, you should use .apply(fahr_to_celsius) not .apply(fahr_to_celsius()). Why? Because the .apply() method will execute and use the function itself in the background when it operates with the data. If we would pass our function including the parentheses, the fahr_to_celsius() function would actually be executed once before the loop with .apply() starts (thus becoming unusable) and that is not what we want.

Our previous command only returned the Series of temperatures to the screen, but naturally we can also store them permanently in a column of our DataFrame (overwriting the old values).

data["TMIN_C"] = data["TMIN_F"].apply(fahr_to_celsius)

A nice thing with .apply() is that we can also apply the function on several columns at once. Below, we also sort the values in descending order based on values in the TMIN_F column to confirm that applying our function really works.

cols = ["TMAX_F", "TMIN_F"]
result = data[cols].apply(fahr_to_celsius)
result.sort_values(by="TMIN_F", ascending=False).head()
TMAX_F TMIN_F
1658 31.666667 23.333333
4595 26.666667 22.777778
1683 29.444444 22.777778
1673 27.777778 22.222222
6077 30.000000 22.222222

You can also directly store the outputs to the DataFrame columns TMAX_C and TMIN_C.

cols = ["TMAX_F", "TMIN_F"]
newcols = ["TMAX_C", "TMIN_C"]
data[newcols] = data[cols].apply(fahr_to_celsius)
data.head()
STATION_ID DATE TMAX_F TMIN_F TMIN_C TMAX_C
0 GHCND:FIE00142226 20051213 40.0 32.0 0.000000 4.444444
1 GHCND:FIE00142226 20051214 35.0 26.0 -3.333333 1.666667
2 GHCND:FIE00142226 20051215 38.0 33.0 0.555556 3.333333
3 GHCND:FIE00142226 20051216 36.0 19.0 -7.222222 2.222222
4 GHCND:FIE00142226 20051217 19.0 15.0 -9.444444 -7.222222

In this section, we showed you a few different ways to iterate over rows in pandas and apply functions. The most important thing is that you understand the logic of how loops work and how you can use your own functions to modify the values in a pandas DataFrame. Whenever you need to loop over your data, we recommend using .apply() as it is typically the most efficient one in terms of execution time. Remember that in most cases you do not need to use loops. Instead you can do calculations in a “vectorized manner” (which is the fastest way) as we learned previously when doing basic calculations in pandas.

String slicing#

We will eventually want to group our data based on month in order to see if the temperatures in June of 2021 were higher than on average (which is the goal in our analysis as you might recall). Currently, the date and time information is stored in the column DATE that has a structure yyyyMMdd. This is a typical timestamp format in which yyyy represents the year in a four digit format, MM is for the month (two digits), and dd is for the day. Let’s have a closer look at the date and time information we have by checking the values in that column and their data type.

data["DATE"].head()

0    20051213
1    20051214
2    20051215
3    20051216
4    20051217
Name: DATE, dtype: int64

data["DATE"].tail()

6816    20240827
6817    20240828
6818    20240829
6819    20240830
6820    20240831
Name: DATE, dtype: int64

The DATE column contains dates for the range of observations in our data set. The timestamp for the first observation is 20051213 (December 13th, 2005) and the timestamp for the latest observation is 20240831 (August 31st, 2024). As we can see, the data type (dtype) of our column seems to be int64 (i.e., the information is stored as integer values).

To proceed with our analysis, we want to aggregate this data on a monthly level. In order to do so, we need to “label” each row of data based on the month when the record was observed. Hence, we need to somehow separate information about the year and month for each row. In practice, we can create a new column (or an index) containing information about the month (including the year but excluding days). There are different ways of achieving this, but here we will take advantage of string slicing which means that we convert the date information into character strings and “cut” the needed information from the string objects. The other option would be to convert the timestamp values into something called datetime objects, but we will learn about those a bit later. Before further processing, we first want to convert the DATE column to character strings for convenience and store those values in a new column called DATE_STR.

data["DATE_STR"] = data["DATE"].astype(str)

If we look at the latest time stamp in the data (20240831), you can see that there is a systematic pattern: YEAR-MONTH-DAY. The first four characters always represent the year and the following two characters represent the month. Because we are interested in understanding monthly averages for different years, we want to slice the year and month values from the timestamp (the first 6 characters) like this:

date = "20240831"
date[0:6]

'202408'

Using this approach, we can slice the correct range of characters from the DATE_STR column using a specific pandas function designed for a Series called .str.slice(). The function has the parameters start and stop, which you can use to specify the positions where the slicing should start and end.

data["YEAR_MONTH"] = data["DATE_STR"].str.slice(start=0, stop=6)
data.head()
STATION_ID DATE TMAX_F TMIN_F TMIN_C TMAX_C DATE_STR YEAR_MONTH
0 GHCND:FIE00142226 20051213 40.0 32.0 0.000000 4.444444 20051213 200512
1 GHCND:FIE00142226 20051214 35.0 26.0 -3.333333 1.666667 20051214 200512
2 GHCND:FIE00142226 20051215 38.0 33.0 0.555556 3.333333 20051215 200512
3 GHCND:FIE00142226 20051216 36.0 19.0 -7.222222 2.222222 20051216 200512
4 GHCND:FIE00142226 20051217 19.0 15.0 -9.444444 -7.222222 20051217 200512

Nice! Now we have “labeled” the rows based on information about the year and month.

Question 3.7#

Create a new column MONTH with information about the month without the year.

Hide code cell content 
# Solution

data["MONTH"] = data["DATE_STR"].str.slice(start=4, stop=6)

Grouping and aggregating data#

Basic logic of grouping a DataFrame using .groupby()#

In the following sections, we want to calculate the average temperature for each month in our data set. Here, we will learn how to use the .groupby() method, which is a handy tool for combining large amounts of data and computing statistics for subgroups. We will use the .groupby() method to calculate the average temperatures for each month in three main steps:

  1. Group the data based on year and month using .groupby()

  2. Calculate the average temperature for each month (i.e., each group)

  3. Store the resulting row into a DataFrame called monthly_data

We have many rows of weather data (n = 6821) to process and our goal is to create an aggregated DataFrame that has only one row per month. We can group the data by month using the .groupby() function by providing the name of the column (or a list of columns) that we want to use as basis for doing the grouping. Let’s group our data based on the unique year and month combinations.

grouped = data.groupby("YEAR_MONTH")

Notice, that it would also be possible to create combinations of years and months “on the fly” if you have them in separate columns. In such case, grouping the data could be done as grouped = data.groupby(['YEAR', 'MONTH']). Let’s explore the new variable grouped.

print(type(grouped))
print(len(grouped))

<class 'pandas.core.groupby.generic.DataFrameGroupBy'>
225

We have a new object with type DataFrameGroupBy with 225 groups. In order to understand what just happened, let’s also check the number of unique year and month combinations in our data.

data["YEAR_MONTH"].nunique()

225

Length of the grouped object should be the same as the number of unique values in the column we used for grouping (YEAR_MONTH). For each unique YEAR_MONTH value, there is a group of data. Let’s explore our grouped data further by checking the “names” of the first five groups. Here, we access the keys of the groups and convert them to a list so that we can slice and print only a few of those to the screen.

list(grouped.groups.keys())[:5]

['200512', '200601', '200602', '200603', '200604']

Let’s check the contents for a group representing December 2005. We can get the values for that month from the grouped object using the .get_group() method.

# Specify a month (as character string)
month = "200512"

# Select the group
group1 = grouped.get_group(month)
group1
STATION_ID DATE TMAX_F TMIN_F TMIN_C TMAX_C DATE_STR YEAR_MONTH MONTH
0 GHCND:FIE00142226 20051213 40.0 32.0 0.000000 4.444444 20051213 200512 12
1 GHCND:FIE00142226 20051214 35.0 26.0 -3.333333 1.666667 20051214 200512 12
2 GHCND:FIE00142226 20051215 38.0 33.0 0.555556 3.333333 20051215 200512 12
3 GHCND:FIE00142226 20051216 36.0 19.0 -7.222222 2.222222 20051216 200512 12
4 GHCND:FIE00142226 20051217 19.0 15.0 -9.444444 -7.222222 20051217 200512 12
5 GHCND:FIE00142226 20051218 18.0 12.0 -11.111111 -7.777778 20051218 200512 12
6 GHCND:FIE00142226 20051219 22.0 11.0 -11.666667 -5.555556 20051219 200512 12
7 GHCND:FIE00142226 20051220 22.0 11.0 -11.666667 -5.555556 20051220 200512 12
8 GHCND:FIE00142226 20051221 23.0 15.0 -9.444444 -5.000000 20051221 200512 12
9 GHCND:FIE00142226 20051222 20.0 13.0 -10.555556 -6.666667 20051222 200512 12
10 GHCND:FIE00142226 20051223 34.0 12.0 -11.111111 1.111111 20051223 200512 12
11 GHCND:FIE00142226 20051224 34.0 31.0 -0.555556 1.111111 20051224 200512 12
12 GHCND:FIE00142226 20051225 32.0 28.0 -2.222222 0.000000 20051225 200512 12
13 GHCND:FIE00142226 20051226 32.0 20.0 -6.666667 0.000000 20051226 200512 12
14 GHCND:FIE00142226 20051227 21.0 18.0 -7.777778 -6.111111 20051227 200512 12
15 GHCND:FIE00142226 20051228 24.0 18.0 -7.777778 -4.444444 20051228 200512 12
16 GHCND:FIE00142226 20051229 27.0 18.0 -7.777778 -2.777778 20051229 200512 12
17 GHCND:FIE00142226 20051230 33.0 27.0 -2.777778 0.555556 20051230 200512 12
18 GHCND:FIE00142226 20051231 36.0 31.0 -0.555556 2.222222 20051231 200512 12

As we can see, a single group contains a DataFrame with values only for that specific month. Let’s check the data type of this group.

type(group1)

pandas.core.frame.DataFrame

So, one group is a pandas DataFrame, which is really useful because it allows us to use all the familiar DataFrame methods for calculating statistics, etc. for this specific group. It is also possible to iterate over the groups in our DataFrameGroupBy object which can be useful if you need to conduct and apply some more complicated sub-tasks for each group. When doing so, it is important to understand that a single group in our DataFrameGroupBy object actually contains not only the actual values but also information about the key that was used to do the grouping. Hence, when iterating we need to assign the key and the values (i.e., the group) to separate variables. Let’s see how we can iterate over the groups and print the key and the data from a single group (again using break to only see the first group).

# Iterate over groups
for key, group in grouped:
    # Print key and group
    print(f"Key:\n{key}")
    print(f"\nFirst rows of data in this group:\n {group.head()}")

    # Stop iteration with break command
    break

Key:
200512

First rows of data in this group:
           STATION_ID      DATE  TMAX_F  TMIN_F    TMIN_C    TMAX_C  DATE_STR  \
0  GHCND:FIE00142226  20051213    40.0    32.0  0.000000  4.444444  20051213   
1  GHCND:FIE00142226  20051214    35.0    26.0 -3.333333  1.666667  20051214   
2  GHCND:FIE00142226  20051215    38.0    33.0  0.555556  3.333333  20051215   
3  GHCND:FIE00142226  20051216    36.0    19.0 -7.222222  2.222222  20051216   
4  GHCND:FIE00142226  20051217    19.0    15.0 -9.444444 -7.222222  20051217   

  YEAR_MONTH MONTH  
0     200512    12  
1     200512    12  
2     200512    12  
3     200512    12  
4     200512    12

Here, we can see that the key contains the name of the group (i.e., the unique value from YEAR_MONTH).

Aggregating data with .groupby()#

We can, for example, calculate the average values for all variables using the statistical functions that we have seen already (e.g., .mean(), .std(), .min(), etc.). To calculate the average temperature for each month, we can use the .mean() function. Let’s calculate the mean for all the weather related data attributes in our group at once.

# Specify the columns that will be part of the calculation
mean_cols = ["TMAX_F", "TMIN_F", "TMAX_C", "TMIN_C"]

# Calculate the mean values all at one go
mean_values = group1[mean_cols].mean()
mean_values

TMAX_F    28.736842
TMIN_F    20.526316
TMAX_C    -1.812865
TMIN_C    -6.374269
dtype: float64

As a result, we get a pandas Series with mean values calculated for all columns in the group. Notice that if you want to convert this Series back into a DataFrame (which can be useful if you want to merge multiple groups, for example), you can use the command .to_frame().T, which first converts the Series into a DataFrame and then transposes the order of the axes (the label names become the column names).

# Convert to DataFrame
mean_values.to_frame().T
TMAX_F TMIN_F TMAX_C TMIN_C
0 28.736842 20.526316 -1.812865 -6.374269

To do a similar aggregation with all the groups in our data set, we can combine the .groupby() function with the aggregation step (such as taking the mean of the given columns), and finally restructure the resulting DataFrame a bit. This might be a bit harder to understand at first, but this is how you would group and aggregate the values.

# The columns that we want to aggregate
mean_cols = ["TMAX_F", "TMIN_F", "TMAX_C", "TMIN_C"]

# Group and aggregate the data with one line
monthly_data = data.groupby("YEAR_MONTH")[mean_cols].mean().reset_index()
monthly_data
YEAR_MONTH TMAX_F TMIN_F TMAX_C TMIN_C
0 200512 28.736842 20.526316 -1.812865 -6.374269
1 200601 29.064516 20.580645 -1.630824 -6.344086
2 200602 22.357143 12.464286 -5.357143 -10.853175
3 200603 28.838710 16.064516 -1.756272 -8.853047
4 200604 44.833333 33.266667 7.129630 0.703704
... ... ... ... ... ...
220 202404 45.566667 33.666667 7.537037 0.925926
221 202405 67.322581 47.870968 19.623656 8.817204
222 202406 72.433333 55.266667 22.462963 12.925926
223 202407 74.354839 60.516129 23.530466 15.842294
224 202408 73.032258 58.193548 22.795699 14.551971

225 rows × 5 columns

As we can see, aggregating the data in this way is a fairly straightforward and fast process requiring merely a single command. So what did we actually do here? We (1) grouped the data, (2) selected specific columns from the result (mean_cols), (3) calculated the mean for all of the selected columns of the groups, and finally (4) reset the index. Resetting the index at the end is not necessary, but by doing it we turn the YEAR_MONTH values into a dedicated column in our data (which would be otherwise be stored as index).

What might not be obvious from this example is the fact that each group is actually iterated over and the aggregation step is repeated for each group. For you to better understand what happens, we will next repeat the same process by iterating over the groups and eventually creating a DataFrame that contains the mean values for all of the weather attributes that we are interested in. In this approach, we will iterate over the groups, calculate the mean values, store the result in a list, and finally merge the aggregated data into a DataFrame called monthly_data.

# Create an empty list for storing the aggregated rows/DataFrames
data_container = []

# The columns that we want to aggregate
mean_cols = ["TMAX_F", "TMIN_F", "TMAX_C", "TMIN_C"]

# Iterate over the groups
for key, group in grouped:
    # Calculate mean
    mean_values = group[mean_cols].mean()

    # Add the "key" (i.e., the date+time information) into the Series
    mean_values["YEAR_MONTH"] = key

    # Convert the Series into a DataFrame and
    # append the aggregated values into a list as a DataFrame
    data_container.append(mean_values.to_frame().T)

# After iterating over all groups, merge the list of DataFrames
monthly_data = pd.concat(data_container)
monthly_data
TMAX_F TMIN_F TMAX_C TMIN_C YEAR_MONTH
0 28.736842 20.526316 -1.812865 -6.374269 200512
0 29.064516 20.580645 -1.630824 -6.344086 200601
0 22.357143 12.464286 -5.357143 -10.853175 200602
0 28.83871 16.064516 -1.756272 -8.853047 200603
0 44.833333 33.266667 7.12963 0.703704 200604
... ... ... ... ... ...
0 45.566667 33.666667 7.537037 0.925926 202404
0 67.322581 47.870968 19.623656 8.817204 202405
0 72.433333 55.266667 22.462963 12.925926 202406
0 74.354839 60.516129 23.530466 15.842294 202407
0 73.032258 58.193548 22.795699 14.551971 202408

225 rows × 5 columns

As a result, we get identical results to those produced by the approach that was done earlier with a single line of code (except for the position of the YEAR_MONTH column).

So which approach should you use? From the performance point of view, we recommend using the first approach (i.e., chaining), which does not require you to use a separate for loop, and is highly efficient. However, this approach might be a bit more difficult to read and comprehend (the loop might be easier). Also, sometimes you want to include additional processing steps within the loop that can be hard accomplish by chaining everything into a single command. Hence, it is useful to know both of these approaches for doing aggregations of the data.

Case study: Detecting warm months#

Now that we have aggregated our data on monthly level, all we need to do is to check which years had the warmest June temperatures. A simple approach is to select all June values from the data and check which group(s) have the highest mean value. Before doing this, let’s separate the month information from our timestamp following the same approach as previously we did when slicing the year-month combination.

monthly_data["MONTH"] = monthly_data["YEAR_MONTH"].str.slice(start=4, stop=6)
monthly_data.head()
TMAX_F TMIN_F TMAX_C TMIN_C YEAR_MONTH MONTH
0 28.736842 20.526316 -1.812865 -6.374269 200512 12
0 29.064516 20.580645 -1.630824 -6.344086 200601 01
0 22.357143 12.464286 -5.357143 -10.853175 200602 02
0 28.83871 16.064516 -1.756272 -8.853047 200603 03
0 44.833333 33.266667 7.12963 0.703704 200604 04

Let’s also make an estimate of the mean monthly temperature TAVG_C as the average of TMIN_C and TMAX_C.

monthly_data["TAVG_C"] = (monthly_data["TMAX_C"] + monthly_data["TMIN_C"]) / 2.0

Now we can select the values for June from our data and store it into a new variable june_data. Then we can check the highest temperature values by sorting the DataFrame by the TAVG_C column in descending order.

june_data = monthly_data.loc[monthly_data["MONTH"] == "06"]
june_data.sort_values(by="TAVG_C", ascending=False).head()
TMAX_F TMIN_F TMAX_C TMIN_C YEAR_MONTH MONTH TAVG_C
0 76.533333 57.8 24.740741 14.333333 202106 06 19.537037
0 72.433333 56.333333 22.462963 13.518519 202006 06 17.990741
0 71.566667 56.766667 21.981481 13.759259 201306 06 17.87037
0 72.433333 55.266667 22.462963 12.925926 202406 06 17.694444
0 72.166667 55.133333 22.314815 12.851852 201906 06 17.583333

By looking at the order of YEAR_MONTH column, we can see that June 2021 is indeed the warmest June on record for the Helsinki Kumpula weather station (as of 2024) based on the estimated average monthly temperatures. Let’s now explore similar temperature data from other weather stations in Finland and see whether June of 2021 has been exceptionally warm in other locations.

Automating the analysis#

Above, we learned how to aggregate data using pandas to calculate average monthly temperatures based on daily weather observations. With these skills, we can now take advantage of one of the most useful aspects of programming: automating processes and repeating analyses for any number of similar data files (assuming the data structure is the same).

Let’s now see how we can repeat the previous data analysis steps for 15 weather stations located in different parts of Finland containing data for up to 116 years (1908-2024). The idea is that we will repeat the process for each input file using a (rather long) for loop. We will use the most efficient alternatives of the previously represented approaches, and finally will store the results in a single DataFrame for all stations. We will learn how to manipulate file paths in Python using the pathlib module and see how we can list our input files in the data directory data/finnish-stations. We will store those paths in the variable file_list so that we can use the file paths easily in the later steps.

Managing and listing filesystem paths#

In Python there are two commonly used approaches to manage and manipulate file paths, namely the os.path sub-module and the newer pathlib module (available since Python 3.4), which we will demonstrate here. The built-in module pathlib provides many useful functions for interacting and manipulating file paths in your operating system. In the following example, we have data in different files in a sub-folder and will learn how to use the Path class from the pathlib library to construct file paths. To start, we will import and use the Path class and see how we can construct a file path by joining a folder path and a file name.

from pathlib import Path

# Initialize the Path
input_folder = Path("data/finnish-stations")

# Join folder path and filename
fp = input_folder / "enontekio-kilpisjarvi.txt"
fp

PosixPath('data/finnish-stations/enontekio-kilpisjarvi.txt')

Above, we first initialized the Path object and stored it in the variable input_folder by passing a relative path as a string indicating the directory where all our files are located. Then we created a complete file path to the file enontekio-kilpisjarvi.txt by adding a forward slash (/) character between the folder and the filename which joins them together (easy!). In this case, our end result is something called a PosixPath, which is a file system path to a given file in the Linux or macOS operating systems. If you would run the same commands on a computer using the Windows operating system, the end result would be a WindowsPath object. Thus, the output depends on which operating system you are using. However, you do not need to worry about this because both types of Paths work exactly the same no matter which operating system you use.

Both the Path object that we stored in the input_folder variable and the PosixPath object that we stored in variable fp are actually quite versatile creatures, and we can do many useful things with them. For instance, we can find the parent folder where the file is located, extract the filename from the full path, test whether the file or directory actually exists, find various properties of the file (such as size of the file or creation time), and so on. Let’s see a few examples below.

fp.parent

PosixPath('data/finnish-stations')

fp.name

'enontekio-kilpisjarvi.txt'

fp.exists()

True

# File properties
size_in_bytes = fp.stat().st_size
creation_time = fp.stat().st_ctime
modified_time = fp.stat().st_mtime
print(
    f"Size (bytes): {size_in_bytes}\nCreated (seconds since Epoch): {creation_time}\nModified (seconds since Epoch): {modified_time}"
)

Size (bytes): 1622542
Created (seconds since Epoch): 1730239456.5983868
Modified (seconds since Epoch): 1730201423.325065

There are also various other methods that you can use from the pathlib module, such as for renaming files (.rename()) or creating directories (.mkdir()). You can see all available methods from the pathlib documentation [3]. One of the most useful tools in pathlib is the ability to list all of the files within a given directory using the .glob() method, which also allows you to add specific search criteria for listing only specific files from a directory.

file_list = list(input_folder.glob("*txt"))

Here, the result is stored in the variable file_list as a list. By default, the .glob() function produces something called a generator which is a “lazy iterator” (a special kind of function that allows you to iterate over items like a list, but without actually storing the data in memory). By enclosing the .glob() search functionality in the list() function, we convert this generator into a normal Python list. Note that we’re using the * character as a wildcard, so any filename that ends with txt will be added to the list of files. Let’s take a look what we produced as a result.

print(f"Number of files in the list: {len(file_list)}")
file_list

Number of files in the list: 15

[PosixPath('data/finnish-stations/kalajoki-ulkokalla.txt'),
 PosixPath('data/finnish-stations/vaasa-airport.txt'),
 PosixPath('data/finnish-stations/turku-airport.txt'),
 PosixPath('data/finnish-stations/enontekio-kilpisjarvi.txt'),
 PosixPath('data/finnish-stations/helsinki-kumpula.txt'),
 PosixPath('data/finnish-stations/parainen-uto.txt'),
 PosixPath('data/finnish-stations/helsinki-kaisaniemi-aws.txt'),
 PosixPath('data/finnish-stations/kuusamo-airport.txt'),
 PosixPath('data/finnish-stations/jyvaskyla-airport.txt'),
 PosixPath('data/finnish-stations/sodankyla-aws.txt'),
 PosixPath('data/finnish-stations/kuopio-ritoniemi.txt'),
 PosixPath('data/finnish-stations/oulu-vihreasaari.txt'),
 PosixPath('data/finnish-stations/tampere-pirkkala-airport.txt'),
 PosixPath('data/finnish-stations/kajaani-airport.txt'),
 PosixPath('data/finnish-stations/hanko-russaro.txt')]

Iterate over input files and repeat the analysis#

At this point we should have all the relevant file paths in the file_list variable, and we can loop over the list using a for loop (again we will break the loop after the first iteration).

for fp in file_list:
    print(fp)
    break

data/finnish-stations/kalajoki-ulkokalla.txt

The data we have in the data files are formatted just like the example file we have used in this section (in fact, it is one of the files in our data directory). Thus, we can easily the file data using the pd.read_csv() function with the same parameter values specified earlier.

data = pd.read_csv(
    fp,
    sep=r"\s+",
    skiprows=[1],
    na_values=["-9999"],
    usecols=["STATION", "DATE", "TMAX", "TMIN"],
)
data.head()
STATION DATE TMAX TMIN
0 GHCND:FIE00145151 19951012 50.0 45.0
1 GHCND:FIE00145151 19951013 49.0 45.0
2 GHCND:FIE00145151 19951014 50.0 41.0
3 GHCND:FIE00145151 19951015 42.0 36.0
4 GHCND:FIE00145151 19951016 48.0 38.0

Now that we have all the file paths for our weather observation data sets in a list, we can start iterating over them and repeat the analysis steps for each file separately. We will keep all the analytical steps inside a loop so that all of them are repeated for the different stations. Finally, we will store the warmest June for each station in a list called results using a regular Python list’s .append() method and merge the list of DataFrames into one using the pd.concat() function.

# A list for storing the result
results = []

# Repeat the analysis steps for each input file:
for fp in file_list:
    # Read the data from text file
    data = pd.read_csv(
        fp,
        sep=r"\s+",
        skiprows=[1],
        na_values=["-9999"],
        usecols=["STATION", "DATE", "TMAX", "TMIN"],
    )

    # Rename the columns
    new_names = {
        "STATION": "STATION_ID",
        "TMAX": "TMAX_F",
        "TMIN": "TMIN_F",
    }
    data = data.rename(columns=new_names)

    # Print info about the current input file
    # This is useful to understand how the process proceeds
    print(
        f"STATION NUMBER: {data.at[0,'STATION_ID']}\tNUMBER OF OBSERVATIONS: {len(data)}"
    )

    # Estimate daily average temperature as average of TMAX and TMIN
    data["TAVG_F"] = (data["TMAX_F"] + data["TMIN_F"]) / 2.0

    # Create column
    data["TAVG_C"] = None

    # Convert temperatures from Fahrenheit to Celsius
    data["TAVG_C"] = data["TAVG_F"].apply(fahr_to_celsius)

    # Convert DATE to string
    data["DATE_STR"] = data["DATE"].astype(str)

    # Parse year and month and convert them to numbers
    data["MONTH"] = data["DATE_STR"].str.slice(start=5, stop=6).astype(int)
    data["YEAR"] = data["DATE_STR"].str.slice(start=0, stop=4).astype(int)

    # Extract observations for the months of June
    june = data[data["MONTH"] == 6]

    # Aggregate the data and get mean values
    columns = ["TAVG_F", "TAVG_C"]
    monthly_mean = june.groupby(by=["YEAR", "MONTH"])[columns].mean().reset_index()

    # Sort the values and take the warmest June
    warmest = monthly_mean.sort_values(by="TAVG_C", ascending=False).head(1)

    # Save data file name in DataFrame
    warmest["FILE"] = fp.name

    # Add to results
    results.append(warmest)

# Merge all the results into a single DataFrame
results = pd.concat(results)

STATION NUMBER: GHCND:FIE00145151	NUMBER OF OBSERVATIONS: 9733
STATION NUMBER: GHCND:FIE00144212	NUMBER OF OBSERVATIONS: 25075
STATION NUMBER: GHCND:FIE00142511	NUMBER OF OBSERVATIONS: 26167
STATION NUMBER: GHCND:FIM00002801	NUMBER OF OBSERVATIONS: 15305
STATION NUMBER: GHCND:FIE00142226	NUMBER OF OBSERVATIONS: 6821
STATION NUMBER: GHCND:FIE00141810	NUMBER OF OBSERVATIONS: 24041
STATION NUMBER: GHCND:FI000000304	NUMBER OF OBSERVATIONS: 26907
STATION NUMBER: GHCND:FIE00146062	NUMBER OF OBSERVATIONS: 20672
STATION NUMBER: GHCND:FI000002401	NUMBER OF OBSERVATIONS: 26963
STATION NUMBER: GHCND:FI000007501	NUMBER OF OBSERVATIONS: 42593
STATION NUMBER: GHCND:FIE00144751	NUMBER OF OBSERVATIONS: 10739
STATION NUMBER: GHCND:FIE00145661	NUMBER OF OBSERVATIONS: 10118
STATION NUMBER: GHCND:FIE00142706	NUMBER OF OBSERVATIONS: 16356
STATION NUMBER: GHCND:FIE00145377	NUMBER OF OBSERVATIONS: 25902
STATION NUMBER: GHCND:FIE00141925	NUMBER OF OBSERVATIONS: 23968

Awesome! Now we have conducted the same analysis for 15 weather stations in Finland and it did not take many lines of code! We were able to follow how the process in real time via the information printed as the data were analyzed (i.e., we did some simple “logging” of the operations). Let’s finally investigate our results.

results
YEAR MONTH TAVG_F TAVG_C FILE
26 2024 6 58.266667 14.592593 kalajoki-ulkokalla.txt
65 2020 6 61.716667 16.509259 vaasa-airport.txt
69 2021 6 64.333333 17.962963 turku-airport.txt
41 2022 6 53.666667 12.037037 enontekio-kilpisjarvi.txt
15 2021 6 67.166667 19.537037 helsinki-kumpula.txt
62 2021 6 62.166667 16.759259 parainen-uto.txt
70 2021 6 66.600000 19.222222 helsinki-kaisaniemi-aws.txt
54 2022 6 61.833333 16.574074 kuusamo-airport.txt
70 2021 6 64.400000 18.000000 jyvaskyla-airport.txt
45 1953 6 60.483333 15.824074 sodankyla-aws.txt
28 2021 6 66.400000 19.111111 kuopio-ritoniemi.txt
16 2013 6 62.300000 16.833333 oulu-vihreasaari.txt
41 2021 6 65.100000 18.388889 tampere-pirkkala-airport.txt
1 1953 6 61.968750 16.649306 kajaani-airport.txt
62 2021 6 63.216667 17.342593 hanko-russaro.txt

Each row in the results represents the warmest June at given station starting between 1908-2005 and ending in 2024. Based on the YEAR column, the warmest June in the majority of weather stations we have used was indeed in 2021. We can confirm this by checking the value counts for the YEAR column.

results["YEAR"].value_counts()

YEAR
2021    8
2022    2
1953    2
2024    1
2020    1
2013    1
Name: count, dtype: int64

It seems at least the start of summer in 2021 was abnormally warm in Finland!

Footnotes#