Pandas 按列和均值散点图对 Excel 数据进行分组

Question

I have a collection of data I am reading out from several Excel files. I can easily read, merge and group the data with pandas. I have two columns of interest in the data, 'Product Type' and 'Test Duration.'

The dataframe containing the data read from the Excel file is called oData.

oDataGroupedByProductType = oData.groupby(['Product Type'])

I have used plotly to make a graph as follows, but plotly does not keep the data private and if I want the data to be private I have to pay. Paying is not an option. How can I make the same graph with pandas and/or matplotlib, but also with the mean for each product type displayed?

Answer 1

As Bound says, you can do it a few lines with stripplot (Example of the seaborn documentation page).

import seaborn as sns
sns.set_style("whitegrid")
tips = sns.load_dataset("tips") # load some sample data
ax = sns.stripplot(x="day", y="total_bill", data=tips)

Answer 2

Suppose you have some dataframe:

In [4]: df.head(20)
Out[4]:
   product      value
0        c   5.155740
1        c   8.983128
2        c   5.150390
3        a   8.379866
4        c   8.094536
5        c   7.464706
6        b   3.690430
7        a   5.547448
8        a   7.709569
9        c   8.398026
10       a   7.317957
11       b   7.821332
12       b   8.815495
13       c   6.646533
14       c   8.239603
15       c   7.585408
16       a   7.946760
17       c   5.276864
18       c   8.793054
19       b  11.573413

You need to have a numeric value for the product to plot it, so quick-and-drity, just make a new column by mapping numeric values:

In [5]: product_map = {p:r for p,r in zip(df['product'].unique(), range(1, df.values.shape[0]+1))}

In [6]: product_map
Out[6]: {'a': 2, 'b': 3, 'c': 1}

Of course, there are many ways you could achieve this...

Now, make a new column:

In [8]: df['product_code'] = df['product'].map(product_map)

In [9]: df.head(20)
Out[9]:
   product      value  product_code
0        c   5.155740             1
1        c   8.983128             1
2        c   5.150390             1
3        a   8.379866             2
4        c   8.094536             1
5        c   7.464706             1
6        b   3.690430             3
7        a   5.547448             2
8        a   7.709569             2
9        c   8.398026             1
10       a   7.317957             2
11       b   7.821332             3
12       b   8.815495             3
13       c   6.646533             1
14       c   8.239603             1
15       c   7.585408             1
16       a   7.946760             2
17       c   5.276864             1
18       c   8.793054             1
19       b  11.573413             3

Now, use the plot helper method in pandas which is basically a wrapper around matplotlib :

In [10]: df.plot(kind='scatter', x = 'product_code', y = 'value')
Out[10]: <matplotlib.axes._subplots.AxesSubplot at 0x12235abe0>

And the output:

Clearly, this was quick and dirty, but it should get you on your way...

Answer 3

In case someone else has a very similar problem and wants to see the end results, I ended up using seaborn, as follows:

import seaborn as sns
import matplotlib.pyplot as plt
...
sns.set_style("whitegrid")
sns.boxplot(x=oData['Product Type'],
          y=oData['Test Duration?'],
          data=oData)
plt.savefig('Test Duration vs. Product Type.png')

The graph came out as follows. For privacy reasons, I have blurred the product labels on the graph.