Practical Business Python

Taking care of business, one python script at a time

Wed 17 December 2014

Combining Data From Multiple Excel Files

Posted by Chris Moffitt in articles   

Introduction

A common task for python and pandas is to automate the process of aggregating data from multiple files and spreadsheets.

This article will walk through the basic flow required to parse multiple Excel files, combine the data, clean it up and analyze it. The combination of python + pandas can be extremely powerful for these activities and can be a very useful alternative to the manual processes or painful VBA scripts frequently used in business settings today.

The Problem

Before, I get into the examples, here is a simple diagram showing the challenges with the common process used in businesses all over the world to consolidate data from multiple Excel files, clean it up and perform some analysis.

Excel file processing

If you’re reading this article, I suspect you have experienced some of the problems shown above. Cutting and pasting data or writing painful VBA code will quickly get old. There has to be a better way!

Python + pandas can be a great alternative that is much more scaleable and powerful.

Excel file processing with pandas

By using a python script, you can develop a more streamlined and repeatable solution to your data processing needs. The rest of this article will show a simple example of how this process works. I hope it will give you ideas of how to apply these tools to your unique situation.

Collecting the Data

If you are interested in following along, here are the excel files and a link to the notebook:

The first step in the process is collecting all the data into one place.

First, import pandas and numpy

import pandas as pd
import numpy as np

Let’s take a look at the files in our input directory, using the convenient shell commands in ipython.

!ls ../in
address-state-example.xlsx  report.xlsx                sample-address-new.xlsx
customer-status.xlsx            sales-feb-2014.xlsx    sample-address-old.xlsx
excel-comp-data.xlsx            sales-jan-2014.xlsx    sample-diff-1.xlsx
my-diff-1.xlsx                  sales-mar-2014.xlsx    sample-diff-2.xlsx
my-diff-2.xlsx                  sample-address-1.xlsx  sample-salesv3.xlsx
my-diff.xlsx                    sample-address-2.xlsx
pricing.xlsx                    sample-address-3.xlsx

There are a lot of files, but we only want to look at the sales .xlsx files.

!ls ../in/sales*.xlsx
../in/sales-feb-2014.xlsx  ../in/sales-jan-2014.xlsx  ../in/sales-mar-2014.xlsx

Use the python glob module to easily list out the files we need.

import glob
glob.glob("../in/sales*.xlsx")
['../in/sales-jan-2014.xlsx',
 '../in/sales-mar-2014.xlsx',
 '../in/sales-feb-2014.xlsx']

This gives us what we need. Let’s import each of our files and combine them into one file. Panda’s concat and append can do this for us. I’m going to use append in this example.

The code snippet below will initialize a blank DataFrame then append all of the individual files into the all_data DataFrame.

all_data = pd.DataFrame()
for f in glob.glob("../in/sales*.xlsx"):
    df = pd.read_excel(f)
    all_data = all_data.append(df,ignore_index=True)

Now we have all the data in our all_data DataFrame. You can use describe to look at it and make sure you data looks good.

all_data.describe()
account number quantity unit price ext price
count 1742.000000 1742.000000 1742.000000 1742.000000
mean 485766.487945 24.319173 54.985454 1349.229392
std 223750.660792 14.502759 26.108490 1094.639319
min 141962.000000 -1.000000 10.030000 -97.160000
25% 257198.000000 12.000000 32.132500 468.592500
50% 527099.000000 25.000000 55.465000 1049.700000
75% 714466.000000 37.000000 77.607500 2074.972500
max 786968.000000 49.000000 99.850000 4824.540000

A lot of this data may not make much sense for this data set but I’m most interested in the count row to make sure the number of data elements makes sense. In this case, I see all the data rows I expect.

all_data.head()
account number name sku quantity unit price ext price date
0 740150 Barton LLC B1-20000 39 86.69 3380.91 2014-01-01 07:21:51
1 714466 Trantow-Barrows S2-77896 -1 63.16 -63.16 2014-01-01 10:00:47
2 218895 Kulas Inc B1-69924 23 90.70 2086.10 2014-01-01 13:24:58
3 307599 Kassulke, Ondricka and Metz S1-65481 41 21.05 863.05 2014-01-01 15:05:22
4 412290 Jerde-Hilpert S2-34077 6 83.21 499.26 2014-01-01 23:26:55

It is not critical in this example but the best practice is to convert the date column to a date time object.

all_data['date'] = pd.to_datetime(all_data['date'])

Combining Data

Now that we have all of the data into one DataFrame, we can do any manipulations the DataFrame supports. In this case, the next thing we want to do is read in another file that contains the customer status by account. You can think of this as a company’s customer segmentation strategy or some other mechanism for identifying their customers.

First, we read in the data.

status = pd.read_excel("../in/customer-status.xlsx")
status
account number name status
0 740150 Barton LLC gold
1 714466 Trantow-Barrows silver
2 218895 Kulas Inc bronze
3 307599 Kassulke, Ondricka and Metz bronze
4 412290 Jerde-Hilpert bronze
5 729833 Koepp Ltd silver
6 146832 Kiehn-Spinka silver
7 688981 Keeling LLC silver
8 786968 Frami, Hills and Schmidt silver
9 239344 Stokes LLC gold
10 672390 Kuhn-Gusikowski silver
11 141962 Herman LLC gold
12 424914 White-Trantow silver
13 527099 Sanford and Sons bronze
14 642753 Pollich LLC bronze
15 257198 Cronin, Oberbrunner and Spencer gold

We want to merge this data with our concatenated data set of sales. Use panda’s merge function and tell it to do a left join which is similar to Excel’s vlookup function.

all_data_st = pd.merge(all_data, status, how='left')
all_data_st.head()
account number name sku quantity unit price ext price date status
0 740150 Barton LLC B1-20000 39 86.69 3380.91 2014-01-01 07:21:51 gold
1 714466 Trantow-Barrows S2-77896 -1 63.16 -63.16 2014-01-01 10:00:47 silver
2 218895 Kulas Inc B1-69924 23 90.70 2086.10 2014-01-01 13:24:58 bronze
3 307599 Kassulke, Ondricka and Metz S1-65481 41 21.05 863.05 2014-01-01 15:05:22 bronze
4 412290 Jerde-Hilpert S2-34077 6 83.21 499.26 2014-01-01 23:26:55 bronze

This looks pretty good but let’s look at a specific account.

all_data_st[all_data_st["account number"]==737550].head()
account number name sku quantity unit price ext price date status
9 737550 Fritsch, Russel and Anderson S2-82423 14 81.92 1146.88 2014-01-03 19:07:37 NaN
14 737550 Fritsch, Russel and Anderson B1-53102 23 71.56 1645.88 2014-01-04 08:57:48 NaN
26 737550 Fritsch, Russel and Anderson B1-53636 42 42.06 1766.52 2014-01-08 00:02:11 NaN
32 737550 Fritsch, Russel and Anderson S1-27722 20 29.54 590.80 2014-01-09 13:20:40 NaN
42 737550 Fritsch, Russel and Anderson S1-93683 22 71.68 1576.96 2014-01-11 23:47:36 NaN

This account number was not in our status file, so we have a bunch of NaN’s. We can decide how we want to handle this situation. For this specific case, let’s label all missing accounts as bronze. Use the fillna function to easily accomplish this on the status column.

all_data_st['status'].fillna('bronze',inplace=True)
all_data_st.head()
account number name sku quantity unit price ext price date status
0 740150 Barton LLC B1-20000 39 86.69 3380.91 2014-01-01 07:21:51 gold
1 714466 Trantow-Barrows S2-77896 -1 63.16 -63.16 2014-01-01 10:00:47 silver
2 218895 Kulas Inc B1-69924 23 90.70 2086.10 2014-01-01 13:24:58 bronze
3 307599 Kassulke, Ondricka and Metz S1-65481 41 21.05 863.05 2014-01-01 15:05:22 bronze
4 412290 Jerde-Hilpert S2-34077 6 83.21 499.26 2014-01-01 23:26:55 bronze

Check the data just to make sure we’re all good.

all_data_st[all_data_st["account number"]==737550].head()
account number name sku quantity unit price ext price date status
9 737550 Fritsch, Russel and Anderson S2-82423 14 81.92 1146.88 2014-01-03 19:07:37 bronze
14 737550 Fritsch, Russel and Anderson B1-53102 23 71.56 1645.88 2014-01-04 08:57:48 bronze
26 737550 Fritsch, Russel and Anderson B1-53636 42 42.06 1766.52 2014-01-08 00:02:11 bronze
32 737550 Fritsch, Russel and Anderson S1-27722 20 29.54 590.80 2014-01-09 13:20:40 bronze
42 737550 Fritsch, Russel and Anderson S1-93683 22 71.68 1576.96 2014-01-11 23:47:36 bronze

Now we have all of the data along with the status column filled in. We can do our normal data manipulations using the full suite of pandas capability.

Using Categories

One of the relatively new functions in pandas is support for categorical data. From the pandas, documentation:

Categoricals are a pandas data type, which correspond to categorical variables in statistics: a variable, which can take on only a limited, and usually fixed, number of possible values (categories; levels in R). Examples are gender, social class, blood types, country affiliations, observation time or ratings via Likert scales.

For our purposes, the status field is a good candidate for a category type.

Version Warning
You must make sure you have a recent version of pandas ( > 0.15) installed for this example to work.
pd.__version__
'0.15.2'

First, we typecast it the column to a category using astype .

all_data_st["status"] = all_data_st["status"].astype("category")

This doesn’t immediately appear to change anything yet.

all_data_st.head()
account number name sku quantity unit price ext price date status
0 740150 Barton LLC B1-20000 39 86.69 3380.91 2014-01-01 07:21:51 gold
1 714466 Trantow-Barrows S2-77896 -1 63.16 -63.16 2014-01-01 10:00:47 silver
2 218895 Kulas Inc B1-69924 23 90.70 2086.10 2014-01-01 13:24:58 bronze
3 307599 Kassulke, Ondricka and Metz S1-65481 41 21.05 863.05 2014-01-01 15:05:22 bronze
4 412290 Jerde-Hilpert S2-34077 6 83.21 499.26 2014-01-01 23:26:55 bronze

Buy you can see that it is a new data type.

all_data_st.dtypes
account number             int64
name                      object
sku                       object
quantity                   int64
unit price               float64
ext price                float64
date              datetime64[ns]
status                  category
dtype: object

Categories get more interesting when you assign order to the categories. Right now, if we call sort on the column, it will sort alphabetically.

all_data_st.sort(columns=["status"]).head()
account number name sku quantity unit price ext price date status
1741 642753 Pollich LLC B1-04202 8 95.86 766.88 2014-02-28 23:47:32 bronze
1232 218895 Kulas Inc S1-06532 29 42.75 1239.75 2014-09-21 11:27:55 bronze
579 527099 Sanford and Sons S1-27722 41 87.86 3602.26 2014-04-14 18:36:11 bronze
580 383080 Will LLC B1-20000 40 51.73 2069.20 2014-04-14 22:44:58 bronze
581 383080 Will LLC S2-10342 15 76.75 1151.25 2014-04-15 02:57:43 bronze

We use set_categories to tell it the order we want to use for this category object. In this case, we use the Olympic medal ordering.

all_data_st["status"].cat.set_categories([ "gold","silver","bronze"],inplace=True)

Now, we can sort it so that gold shows on top.

all_data_st.sort(columns=["status"]).head()
account number name sku quantity unit price ext price date status
0 740150 Barton LLC B1-20000 39 86.69 3380.91 2014-01-01 07:21:51 gold
1193 257198 Cronin, Oberbrunner and Spencer S2-82423 23 52.90 1216.70 2014-09-09 03:06:30 gold
1194 141962 Herman LLC B1-86481 45 52.78 2375.10 2014-09-09 11:49:45 gold
1195 257198 Cronin, Oberbrunner and Spencer B1-50809 30 51.96 1558.80 2014-09-09 21:14:31 gold
1197 239344 Stokes LLC B1-65551 43 15.24 655.32 2014-09-10 11:10:02 gold

Analyze Data

The final step in the process is to analyze the data. Now that it is consolidated and cleaned, we can see if there are any insights to be learned.

all_data_st["status"].describe()
count       1742
unique         3
top       bronze
freq         764
Name: status, dtype: object

For instance, if you want to take a quick look at how your top tier customers are performaing compared to the bottom. Use groupby to get the average of the values.

all_data_st.groupby(["status"])["quantity","unit price","ext price"].mean()
quantity unit price ext price
status
gold 24.680723 52.431205 1325.566867
silver 23.814241 55.724241 1339.477539
bronze 24.589005 55.470733 1367.757736

Of course, you can run multiple aggregation functions on the data to get really useful information

all_data_st.groupby(["status"])["quantity","unit price","ext price"].agg([np.sum,np.mean, np.std])
quantity unit price ext price
sum mean std sum mean std sum mean std
status
gold 8194 24.680723 14.478670 17407.16 52.431205 26.244516 440088.20 1325.566867 1074.564373
silver 15384 23.814241 14.519044 35997.86 55.724241 26.053569 865302.49 1339.477539 1094.908529
bronze 18786 24.589005 14.506515 42379.64 55.470733 26.062149 1044966.91 1367.757736 1104.129089

So, what does this tell you? Well, the data is completely random but my first observation is that we sell more units to our bronze customers than gold. Even when you look at the total dollar value associated with bronze vs. gold, it looks odd that we sell more to bronze customers than gold.

Maybe we should look at how many bronze customers we have and see what is going on?

What I plan to do is filter out the unique accounts and see how many gold, silver and bronze customers there are.

I’m purposely stringing a lot of commands together which is not necessarily best practice but does show how powerful pandas can be. Feel free to review my previous article here and here to understand it better. Play with this command yourself to understand how the commands interact.

all_data_st.drop_duplicates(subset=["account number","name"]).ix[:,[0,1,7]].groupby(["status"])["name"].count()
status
gold      4
silver    7
bronze    9
Name: name, dtype: int64

Ok. This makes a little more sense. We see that we have 9 bronze customers and only 4 customers. That is probably why the volumes are so skewed towards our bronze customers. This result makes sense given the fact that we defaulted to bronze for many of our customers. Maybe we should reclassify some of them? Obviously this data is fake but hopefully this shows how you can use these tools to quickly analyze your own data.

Conclusion

This example only covered the aggregation of 4 simple Excel files containing random data. However the principles can be applied to much larger data sets yet you can keep the code base very manageable. Additionally, you have the full power of python at your fingertips so you can do much more than just simply manipulate the data.

I encourage you to try some of these concepts out on your scenarios and see if you can find a way to automate that painful Excel task that hangs over your head every day, week or month.

Good luck!


 
       Vote on Hacker News          

Comments