It is very convenient to make a data analysis crosstab with one line of code

In the last article we learned that Pandas Module pivot_table() Function can be used to make PivotTables , Today, I'd like to introduce Pandas Another function in the module corsstab(), We can make a crosstab by calling this function , Let's take a look at the main processes and steps .

Module import and data reading

Then we'll follow the routine , First, import the module and read the data set to be used , The reference is still the data set from which the PivotTable report was made

import pandas as pd

def load_data():
    return pd.read_csv('coffee_sales.csv', parse_dates=['order_date'])

Here, I make up a function by customizing it , Then read the data by calling this function , In actual work, everyone can operate according to their own preferences

df = load_data()
df.head()

output

a master hand 's first small display

Crosstab is a special pivot table used to count grouping frequency . Simply speaking , It is to form a new one with the non repeating elements in two or more columns DataFrame, The partial value of the intersection of rows and columns of new data is the number of their combination in the original data , Let's start with a simple example , The code is as follows

pd.crosstab(index = df['region'], columns = df['product_category'])

output

In the row direction, it represents different regions , In the column direction, different coffee varieties are represented , The results show the summary data of different coffee varieties in different regions ,

df[(df["region"] == "Central")&(df["product_category"] == "Tea")].shape[0]

output

336

For example, we screened the data that the region is the central region and the variety is tea , The results are 336 Data , Consistent with the results in the crosstab ,

We can change the names of column names and row indexes , By calling rownames Parameters and colnames Parameters , The code is as follows

pd.crosstab(
    index = df['region'], 
    columns = df['product_category'], 
    rownames=['US Region'], 
    colnames=['Product Category']
)

output

In addition to the category of coffee , We also want to know the sales data of different varieties of coffee between wholesale and retail , That's how it works

pd.crosstab(
    index = df['region'], 
    columns = [df['product_category'], df['market']]
)

output

Or is it

pd.crosstab(
    index = df['region'], 
    columns = [df['product_category'], df['market']],
    rownames=['US Region'], 
    colnames=['Product Category', 'Market']
)

output

Output DataFrame The columns in the dataset have two levels , At the top is the kind of coffee , Then there are different markets on the second floor , Of course, we can also add multiple levels of indexes in the row direction , The code is as follows

pd.crosstab(
    index = [df['region'], df['market']], 
    columns = df['product_category']
)

output

Advanced operation

and pd.pivot_table() The function is the same , We can also call the margin Parameters to add up the integrated data , The code is as follows

pd.crosstab(index = df['region'],
            columns = df['product_category'],
            margins = True)

output

We can also specify the column name of the column ,

pd.crosstab(
    index = df['region'],
    columns = df['product_category'], 
    margins = True, 
    margins_name = 'Subtotals'
)

output

There are also parameters normalize Used to normalize all values divided by the sum of the values

pd.crosstab(index = df['region'], 
            columns = df['product_category'],
            normalize = True)

output

We start from the aesthetic point of view , Want to keep two decimal places , The code is as follows

pd.crosstab(
    index = df['region'], 
    columns = df['product_category'], 
    normalize = True
).style.format('{:.2%}')

output

Between and margin Parameters , Putting all the results together equals 100%, The code is as follows

pd.crosstab(
    index = df['region'], 
    columns = df['product_category'], 
    margins = True, 
    normalize = True
).style.format('{:.2%}')

output

Further derivative

Finally, values as well as aggfunc Two parameters , among aggfunc Parameters specifically refer to the specified aggregate function , For example, the average 、 Statistical methods such as summation and median , Yes value The column of the continuity variable specified by the parameter ,

df.info()

output

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 4248 entries, 0 to 4247
Data columns (total 9 columns):
 #   Column            Non-Null Count  Dtype         
---  ------            --------------  -----         
 0   order_date        4248 non-null   datetime64[ns]
 1   market            4248 non-null   object        
 2   region            4248 non-null   object        
 3   product_category  4248 non-null   object        
 4   product           4248 non-null   object        
 5   cost              4248 non-null   int64         
 6   inventory         4248 non-null   int64         
 7   net_profit        4248 non-null   int64         
 8   sales             4248 non-null   int64         
dtypes: datetime64[ns](1), int64(4), object(4)
memory usage: 298.8+ KB

The current dataset “market”、“region”、“product_category”、“product” Four columns are discrete variables , But there is “cost”、“inventory”、“net_profit”、“sales” Four columns are continuous variables , They represent the cost 、 stock 、 Net profit and sales , We want to target different regions 、 Average the cost of different types of coffee , So here's the code

pd.crosstab(
    index = df['region'], 
    columns = df['product_category'], 
    values = df['cost'],
    aggfunc = 'mean'
)

output

If we want to keep two decimal places for the calculated result , The code is as follows

pd.crosstab(
    index = df['region'], 
    columns = df['product_category'], 
    values = df['cost'],
    aggfunc = 'mean'
).round(2)

output

Of course, if there are missing values , We can also replace it with other values to deal with , The code is as follows

pd.crosstab(
    index = df['region'], 
    columns = df['product_category'], 
    values = df['cost'],
    aggfunc = 'mean',
).fillna(0)

output

NO.1

Previous recommendation

Historical articles

【 Fundamentals of machine learning 】 Various gradient descent optimization algorithms are reviewed and summarized

a line Pandas Code to make a PivotTable , Too cattle

Detailed explanation Python In the middle of pip Common commands

【 Data analysis and entertainment gossip 】 from Python Explore Wang Xinling's traffic password in the visual chart

Share 、 Collection 、 give the thumbs-up 、 I'm looking at the arrangement ？