Python pandas excel cell

To obtain nice column names instead of defaults like 'Unnamed: 1' use the names parameter of pd.read_excel. Mutatis mutandis, try replacing

with pd.ExcelFile(inputFile,
                  sheetname=['pnl1 Data ','pnl2 Data','pnl3 Data','pnl4 Data']) as xlsx:
    df1 = pd.read_excel(xlsx, 'pnl1 Data ',skiprows=9, parse_cols="B:H", keep_default_na='FALSE', na_values=['NULL'])#assign column headers
    df2 = pd.read_excel(xlsx, 'pnl2 Data', skiprows=9, parse_cols="B:H", keep_default_na='FALSE', na_values=['NULL'])
    df3 = pd.read_excel(xlsx, 'pnl3 Data', skiprows=9, parse_cols="B:H", keep_default_na='FALSE', na_values=['NULL'])
    df4 = pd.read_excel(xlsx, 'pnl4 Data', skiprows=9, parse_cols="B:H", keep_default_na='FALSE', na_values=['NULL'])

with

sheets = ['pnl1 Data','pnl2 Data','pnl3 Data','pnl4 Data']
df = pd.read_excel(inputFile, sheetname=sheets, skiprows=9, parse_cols="B:H", 
                   names=list('BCDEFG'))
df = {i: df[sheet] for i, sheet in enumerate(sheets, 1)}

This will make df a dict, whose keys are sheet numbers and whose values are
DataFrames. The DataFrames will have colum names B through G, roughly like
the original Excel file.

Thus, instead of referring to numbered variables df1, …, df4 (generally, a bad idea), you’ll have all the DataFrames in the dict df and will be able to access them by numeric indexing: df[1], …, df[4]. Sheet pnl3 Data, for example, would be accessed as df[3].

To access the seventh row, B column value of sheet 'pnl1 Data' of you could then use:

g_int_c = str(df[1].loc[6, 'B'])

---

For example,

import pandas as pd
try: from cStringIO import StringIO         # for Python2
except ImportError: from io import StringIO # for Python3
import textwrap
df1 = pd.read_csv(StringIO(textwrap.dedent("""
          ,,,
          0,1,2,3
          1,4,5,6
          7,8,9,10""")))
df2 = pd.read_csv(StringIO(textwrap.dedent("""
          ,,,
          0,NULL,2,3
          1,4,NULL,NULL""")), converters={i:str for i in range(4)})

sheets = ['pnl1 Data','pnl2 Data']

writer = pd.ExcelWriter('/tmp/output.xlsx')
for df, sheet in zip([df1, df2], sheets):
    print(df)
    #   Unnamed: 0 Unnamed: 1 Unnamed: 2 Unnamed: 3
    # 0          0       NULL          2          3
    # 1          1          4       NULL       NULL
    df.to_excel(writer, sheet)
writer.save()

df = pd.read_excel('/tmp/output.xlsx', sheetname=sheets, names=list('ABCD'), parse_cols="A:E")
df = {i: df[sheet] for i, sheet in enumerate(sheets, 1)}

for key, dfi in df.items():
    print(dfi)
    #    A  B  C   D
    # 0  0  1  2   3
    # 1  1  4  5   6
    # 2  7  8  9  10
    #    A    B    C    D
    # 0  0  NaN  2.0  3.0
    # 1  1  4.0  NaN  NaN

print(df[1].loc[1, 'B'])
# 4

Why learn to work with Excel with Python? Excel is one of the most popular and widely-used data tools; it’s hard to find an organization that doesn’t work with it in some way. From analysts, to sales VPs, to CEOs, various professionals use Excel for both quick stats and serious data crunching.

With Excel being so pervasive, data professionals must be familiar with it. Working with data in Python or R offers serious advantages over Excel’s UI, so finding a way to work with Excel using code is critical. Thankfully, there’s a great tool already out there for using Excel with Python called pandas.

Pandas has excellent methods for reading all kinds of data from Excel files. You can also export your results from pandas back to Excel, if that’s preferred by your intended audience. Pandas is great for other routine data analysis tasks, such as:

• quick Exploratory Data Analysis (EDA)
• drawing attractive plots
• feeding data into machine learning tools like scikit-learn
• building machine learning models on your data
• taking cleaned and processed data to any number of data tools

Pandas is better at automating data processing tasks than Excel, including processing Excel files.

In this tutorial, we are going to show you how to work with Excel files in pandas. We will cover the following concepts.

• setting up your computer with the necessary software
• reading in data from Excel files into pandas
• data exploration in pandas
• visualizing data in pandas using the matplotlib visualization library
• manipulating and reshaping data in pandas
• moving data from pandas into Excel

Note that this tutorial does not provide a deep dive into pandas. To explore pandas more, check out our course.

System Prerequisites

We will use Python 3 and Jupyter Notebook to demonstrate the code in this tutorial.In addition to Python and Jupyter Notebook, you will need the following Python modules:

• matplotlib — data visualization
• NumPy — numerical data functionality
• OpenPyXL — read/write Excel 2010 xlsx/xlsm files
• pandas — data import, clean-up, exploration, and analysis
• xlrd — read Excel data
• xlwt — write to Excel
• XlsxWriter — write to Excel (xlsx) files

There are multiple ways to get set up with all the modules. We cover three of the most common scenarios below.

• If you have Python installed via Anaconda package manager, you can install the required modules using the command conda install. For example, to install pandas, you would execute the command — conda install pandas.
• If you already have a regular, non-Anaconda Python installed on the computer, you can install the required modules using pip. Open your command line program and execute command pip install <module name> to install a module. You should replace <module name> with the actual name of the module you are trying to install. For example, to install pandas, you would execute command — pip install pandas.
• If you don’t have Python already installed, you should get it through the Anaconda package manager. Anaconda provides installers for Windows, Mac, and Linux Computers. If you choose the full installer, you will get all the modules you need, along with Python and pandas within a single package. This is the easiest and fastest way to get started.

The Data Set

In this tutorial, we will use a multi-sheet Excel file we created from Kaggle’s IMDB Scores data. You can download the file here.

Our Excel file has three sheets: ‘1900s,’ ‘2000s,’ and ‘2010s.’ Each sheet has data for movies from those years.

We will use this data set to find the ratings distribution for the movies, visualize movies with highest ratings and net earnings and calculate statistical information about the movies. We will be analyzing and exploring this data using Python and pandas, thus demonstrating pandas capabilities for working with Excel data in Python.

Read data from the Excel file

We need to first import the data from the Excel file into pandas. To do that, we start by importing the pandas module.

import pandas as pd

We then use the pandas’ read_excel method to read in data from the Excel file. The easiest way to call this method is to pass the file name. If no sheet name is specified then it will read the first sheet in the index (as shown below).

excel_file = 'movies.xls'
movies = pd.read_excel(excel_file)

Here, the read_excel method read the data from the Excel file into a pandas DataFrame object. Pandas defaults to storing data in DataFrames. We then stored this DataFrame into a variable called movies.

Pandas has a built-in DataFrame.head() method that we can use to easily display the first few rows of our DataFrame. If no argument is passed, it will display first five rows. If a number is passed, it will display the equal number of rows from the top.

movies.head()

5 rows × 25 columns

Excel files quite often have multiple sheets and the ability to read a specific sheet or all of them is very important. To make this easy, the pandas read_excel method takes an argument called sheetname that tells pandas which sheet to read in the data from. For this, you can either use the sheet name or the sheet number. Sheet numbers start with zero. If the sheetname argument is not given, it defaults to zero and pandas will import the first sheet.

By default, pandas will automatically assign a numeric index or row label starting with zero. You may want to leave the default index as such if your data doesn’t have a column with unique values that can serve as a better index. In case there is a column that you feel would serve as a better index, you can override the default behavior by setting index_col property to a column. It takes a numeric value for setting a single column as index or a list of numeric values for creating a multi-index.

In the below code, we are choosing the first column, ‘Title’, as index (index=0) by passing zero to the index_col argument.

movies_sheet1 = pd.read_excel(excel_file, sheetname=0, index_col=0)
movies_sheet1.head()

5 rows × 24 columns

As you noticed above, our Excel data file has three sheets. We already read the first sheet in a DataFrame above. Now, using the same syntax, we will read in rest of the two sheets too.

movies_sheet2 = pd.read_excel(excel_file, sheetname=1, index_col=0)
movies_sheet2.head()

5 rows × 24 columns

movies_sheet3 = pd.read_excel(excel_file, sheetname=2, index_col=0)
movies_sheet3.head()

5 rows × 24 columns

Since all the three sheets have similar data but for different recordsmovies, we will create a single DataFrame from all the three DataFrames we created above. We will use the pandas concat method for this and pass in the names of the three DataFrames we just created and assign the results to a new DataFrame object, movies. By keeping the DataFrame name same as before, we are over-writing the previously created DataFrame.

movies = pd.concat([movies_sheet1, movies_sheet2, movies_sheet3])

We can check if this concatenation by checking the number of rows in the combined DataFrame by calling the method shape on it that will give us the number of rows and columns.

movies.shape

(5042, 24)

Using the ExcelFile class to read multiple sheets

We can also use the ExcelFile class to work with multiple sheets from the same Excel file. We first wrap the Excel file using ExcelFile and then pass it to read_excel method.

xlsx = pd.ExcelFile(excel_file)
movies_sheets = []
for sheet in xlsx.sheet_names:
   movies_sheets.append(xlsx.parse(sheet))
movies = pd.concat(movies_sheets)

If you are reading an Excel file with a lot of sheets and are creating a lot of DataFrames, ExcelFile is more convenient and efficient in comparison to read_excel. With ExcelFile, you only need to pass the Excel file once, and then you can use it to get the DataFrames. When using read_excel, you pass the Excel file every time and hence the file is loaded again for every sheet. This can be a huge performance drag if the Excel file has many sheets with a large number of rows.

Exploring the data

Now that we have read in the movies data set from our Excel file, we can start exploring it using pandas. A pandas DataFrame stores the data in a tabular format, just like the way Excel displays the data in a sheet. Pandas has a lot of built-in methods to explore the DataFrame we created from the Excel file we just read in.

We already introduced the method head in the previous section that displays few rows from the top from the DataFrame. Let’s look at few more methods that come in handy while exploring the data set.

We can use the shape method to find out the number of rows and columns for the DataFrame.

movies.shape

(5042, 25)

This tells us our Excel file has 5042 records and 25 columns or observations. This can be useful in reporting the number of records and columns and comparing that with the source data set.

We can use the tail method to view the bottom rows. If no parameter is passed, only the bottom five rows are returned.

movies.tail()

5 rows × 25 columns

In Excel, you’re able to sort a sheet based on the values in one or more columns. In pandas, you can do the same thing with the sort_values method. For example, let’s sort our movies DataFrame based on the Gross Earnings column.

sorted_by_gross = movies.sort_values(['Gross Earnings'], ascending=False)

Since we have the data sorted by values in a column, we can do few interesting things with it. For example, we can display the top 10 movies by Gross Earnings.

sorted_by_gross["Gross Earnings"].head(10)

1867 760505847.0
1027 658672302.0
1263 652177271.0
610 623279547.0
611 623279547.0
1774 533316061.0
1281 474544677.0
226 460935665.0
1183 458991599.0
618 448130642.0
Name: Gross Earnings, dtype: float64

We can also create a plot for the top 10 movies by Gross Earnings. Pandas makes it easy to visualize your data with plots and charts through matplotlib, a popular data visualization library. With a couple lines of code, you can start plotting. Moreover, matplotlib plots work well inside Jupyter Notebooks since you can displace the plots right under the code.

First, we import the matplotlib module and set matplotlib to display the plots right in the Jupyter Notebook.

import matplotlib.pyplot as plt%matplotlib inline

We will draw a bar plot where each bar will represent one of the top 10 movies. We can do this by calling the plot method and setting the argument kind to barh. This tells matplotlib to draw a horizontal bar plot.

sorted_by_gross['Gross Earnings'].head(10).plot(kind="barh")
plt.show()

Let’s create a histogram of IMDB Scores to check the distribution of IMDB Scores across all movies. Histograms are a good way to visualize the distribution of a data set. We use the plot method on the IMDB Scores series from our movies DataFrame and pass it the argument.

movies['IMDB Score'].plot(kind="hist")
plt.show()

This data visualization suggests that most of the IMDB Scores fall between six and eight.

Getting statistical information about the data

Pandas has some very handy methods to look at the statistical data about our data set. For example, we can use the describe method to get a statistical summary of the data set.

movies.describe()

The describe method displays below information for each of the columns.

• the count or number of values
• mean
• standard deviation
• minimum, maximum
• 25%, 50%, and 75% quantile

Please note that this information will be calculated only for the numeric values.

We can also use the corresponding method to access this information one at a time. For example, to get the mean of a particular column, you can use the mean method on that column.

movies["Gross Earnings"].mean()

48468407.526809327

Just like mean, there are methods available for each of the statistical information we want to access. You can read about these methods in our free pandas cheat sheet.

Reading files with no header and skipping records

Earlier in this tutorial, we saw some ways to read a particular kind of Excel file that had headers and no rows that needed skipping. Sometimes, the Excel sheet doesn’t have any header row. For such instances, you can tell pandas not to consider the first row as header or columns names. And If the Excel sheet’s first few rows contain data that should not be read in, you can ask the read_excel method to skip a certain number of rows, starting from the top.

For example, look at the top few rows of this Excel file.

This file obviously has no header and first four rows are not actual records and hence should not be read in. We can tell read_excel there is no header by setting argument header to None and we can skip first four rows by setting argument skiprows to four.

movies_skip_rows = pd.read_excel("movies-no-header-skip-rows.xls", header=None, skiprows=4)
movies_skip_rows.head(5)

5 rows × 25 columns

We skipped four rows from the sheet and used none of the rows as the header. Also, notice that one can combine different options in a single read statement. To skip rows at the bottom of the sheet, you can use option skip_footer, which works just like skiprows, the only difference being the rows are counted from the bottom upwards.

The column names in the previous DataFrame are numeric and were allotted as default by the pandas. We can rename the column names to descriptive ones by calling the method columns on the DataFrame and passing the column names as a list.

movies_skip_rows.columns = ['Title', 'Year', 'Genres', 'Language', 'Country', 'Content Rating', 'Duration', 'Aspect Ratio', 'Budget', 'Gross Earnings', 'Director', 'Actor 1', 'Actor 2', 'Actor 3', 'Facebook Likes - Director', 'Facebook Likes - Actor 1', 'Facebook Likes - Actor 2', 'Facebook Likes - Actor 3', 'Facebook Likes - cast Total', 'Facebook likes - Movie', 'Facenumber in posters', 'User Votes', 'Reviews by Users', 'Reviews by Crtiics', 'IMDB Score']
movies_skip_rows.head()

5 rows × 25 columns

Now that we have seen how to read a subset of rows from the Excel file, we can learn how to read a subset of columns.

Reading a subset of columns

Although read_excel defaults to reading and importing all columns, you can choose to import only certain columns. By passing parse_cols=6, we are telling the read_excel method to read only the first columns till index six or first seven columns (the first column being indexed zero).

movies_subset_columns = pd.read_excel(excel_file, parse_cols=6)
movies_subset_columns.head()

Alternatively, you can pass in a list of numbers, which will let you import columns at particular indexes.

Applying formulas on the columns

One of the much-used features of Excel is to apply formulas to create new columns from existing column values. In our Excel file, we have Gross Earnings and Budget columns. We can get Net earnings by subtracting Budget from Gross earnings. We could then apply this formula in the Excel file to all the rows. We can do this in pandas also as shown below.

movies["Net Earnings"] = movies["Gross Earnings"] - movies["Budget"]

Above, we used pandas to create a new column called Net Earnings, and populated it with the difference of Gross Earnings and Budget. It’s worth noting the difference here in how formulas are treated in Excel versus pandas. In Excel, a formula lives in the cell and updates when the data changes — with Python, the calculations happen and the values are stored — if Gross Earnings for one movie was manually changed, Net Earnings won’t be updated.

Let’s use the sort_values method to sort the data by the new column we created and visualize the top 10 movies by Net Earnings.

sorted_movies = movies[['Net Earnings']].sort_values(['Net Earnings'], ascending=[False])sorted_movies.head(10)['Net Earnings'].plot.barh()
plt.show()

Pivot Table in pandas

Advanced Excel users also often use pivot tables. A pivot table summarizes the data of another table by grouping the data on an index and applying operations such as sorting, summing, or averaging. You can use this feature in pandas too.

We need to first identify the column or columns that will serve as the index, and the column(s) on which the summarizing formula will be applied. Let’s start small, by choosing Year as the index column and Gross Earnings as the summarization column and creating a separate DataFrame from this data.

movies_subset = movies[['Year', 'Gross Earnings']]
movies_subset.head()

We now call pivot_table on this subset of data. The method pivot_table takes a parameter index. As mentioned, we want to use Year as the index.

earnings_by_year = movies_subset.pivot_table(index=['Year'])
earnings_by_year.head()

This gave us a pivot table with grouping on Year and summarization on the sum of Gross Earnings. Notice, we didn’t need to specify Gross Earnings column explicitly as pandas automatically identified it the values on which summarization should be applied.

We can use this pivot table to create some data visualizations. We can call the plot method on the DataFrame to create a line plot and call the show method to display the plot in the notebook.

earnings_by_year.plot()
plt.show()

We saw how to pivot with a single column as the index. Things will get more interesting if we can use multiple columns. Let’s create another DataFrame subset but this time we will choose the columns, Country, Language and Gross Earnings.

movies_subset = movies[['Country', 'Language', 'Gross Earnings']]
movies_subset.head()

We will use columns Country and Language as the index for the pivot table. We will use Gross Earnings as summarization table, however, we do not need to specify this explicitly as we saw earlier.

earnings_by_co_lang = movies_subset.pivot_table(index=['Country', 'Language'])
earnings_by_co_lang.head()

Let’s visualize this pivot table with a bar plot. Since there are still few hundred records in this pivot table, we will plot just a few of them.

earnings_by_co_lang.head(20).plot(kind='bar', figsize=(20,8))
plt.show()

Exporting the results to Excel

If you’re going to be working with colleagues who use Excel, saving Excel files out of pandas is important. You can export or write a pandas DataFrame to an Excel file using pandas to_excel method. Pandas uses the xlwt Python module internally for writing to Excel files. The to_excel method is called on the DataFrame we want to export.We also need to pass a filename to which this DataFrame will be written.

movies.to_excel('output.xlsx')

By default, the index is also saved to the output file. However, sometimes the index doesn’t provide any useful information. For example, the movies DataFrame has a numeric auto-increment index, that was not part of the original Excel data.

movies.head()

5 rows × 26 columns

You can choose to skip the index by passing along index-False.

movies.to_excel('output.xlsx', index=False)

We need to be able to make our output files look nice before we can send it out to our co-workers. We can use pandas ExcelWriter class along with the XlsxWriter Python module to apply the formatting.

We can do use these advanced output options by creating a ExcelWriter object and use this object to write to the EXcel file.

writer = pd.ExcelWriter('output.xlsx', engine='xlsxwriter')
movies.to_excel(writer, index=False, sheet_name='report')
workbook = writer.bookworksheet = writer.sheets['report']

We can apply customizations by calling add_format on the workbook we are writing to. Here we are setting header format as bold.

header_fmt = workbook.add_format({'bold': True})
worksheet.set_row(0, None, header_fmt)

Finally, we save the output file by calling the method save on the writer object.

writer.save()

As an example, we saved the data with column headers set as bold. And the saved file looks like the image below.

Like this, one can use XlsxWriter to apply various formatting to the output Excel file.

Conclusion

Pandas is not a replacement for Excel. Both tools have their place in the data analysis workflow and can be very great companion tools. As we demonstrated, pandas can do a lot of complex data analysis and manipulations, which depending on your need and expertise, can go beyond what you can achieve if you are just using Excel. One of the major benefits of using Python and pandas over Excel is that it helps you automate Excel file processing by writing scripts and integrating with your automated data workflow. Pandas also has excellent methods for reading all kinds of data from Excel files. You can export your results from pandas back to Excel too if that’s preferred by your intended audience.

On the other hand, Excel is a such a widely used data tool, it’s not a wise to ignore it. Acquiring expertise in both pandas and Excel and making them work together gives you skills that can help you stand out in your organization.

If you’d like to learn more about this topic, check out Dataquest’s interactive Pandas and NumPy Fundamentals course, and our Data Analyst in Python, and Data Scientist in Python paths that will help you become job-ready in around 6 months.

Время на прочтение
5 мин

Количество просмотров 63K

Excel — это чрезвычайно распространённый инструмент для анализа данных. С ним легко научиться работать, есть он практически на каждом компьютере, а тот, кто его освоил, может с его помощью решать довольно сложные задачи. Python часто считают инструментом, возможности которого практически безграничны, но который освоить сложнее, чем Excel. Автор материала, перевод которого мы сегодня публикуем, хочет рассказать о решении с помощью Python трёх задач, которые обычно решают в Excel. Эта статья представляет собой нечто вроде введения в Python для тех, кто хорошо знает Excel.

Загрузка данных

Начнём с импорта Python-библиотеки pandas и с загрузки в датафреймы данных, которые хранятся на листах sales и states книги Excel. Такие же имена мы дадим и соответствующим датафреймам.

import pandas as pd
sales = pd.read_excel('https://github.com/datagy/mediumdata/raw/master/pythonexcel.xlsx', sheet_name = 'sales')
states = pd.read_excel('https://github.com/datagy/mediumdata/raw/master/pythonexcel.xlsx', sheet_name = 'states')

Теперь воспользуемся методом .head() датафрейма sales для того чтобы вывести элементы, находящиеся в начале датафрейма:

print(sales.head())

Сравним то, что будет выведено, с тем, что можно видеть в Excel.

Сравнение внешнего вида данных, выводимых в Excel, с внешним видом данных, выводимых из датафрейма pandas

Тут можно видеть, что результаты визуализации данных из датафрейма очень похожи на то, что можно видеть в Excel. Но тут имеются и некоторые очень важные различия:

• Нумерация строк в Excel начинается с 1, а в pandas номер (индекс) первой строки равняется 0.
• В Excel столбцы имеют буквенные обозначения, начинающиеся с буквы A, а в pandas названия столбцов соответствуют именам соответствующих переменных.

Продолжим исследование возможностей pandas, позволяющих решать задачи, которые обычно решают в Excel.

Реализация возможностей Excel-функции IF в Python

В Excel существует очень удобная функция IF, которая позволяет, например, записать что-либо в ячейку, основываясь на проверке того, что находится в другой ячейке. Предположим, нужно создать в Excel новый столбец, ячейки которого будут сообщать нам о том, превышают ли 500 значения, записанные в соответствующие ячейки столбца B. В Excel такому столбцу (в нашем случае это столбец E) можно назначить заголовок MoreThan500, записав соответствующий текст в ячейку E1. После этого, в ячейке E2, можно ввести следующее:

=IF([@Sales]>500, "Yes", "No")

Использование функции IF в Excel

Для того чтобы сделать то же самое с использованием pandas, можно воспользоваться списковым включением (list comprehension):

sales['MoreThan500'] = ['Yes' if x > 500 else 'No' for x in sales['Sales']]

Списковые включения в Python: если текущее значение больше 500 — в список попадает Yes, в противном случае — No

Списковые включения — это отличное средство для решения подобных задач, позволяющее упростить код за счёт уменьшения потребности в сложных конструкциях вида if/else. Ту же задачу можно решить и с помощью if/else, но предложенный подход экономит время и делает код немного чище. Подробности о списковых включениях можно найти здесь.

Реализация возможностей Excel-функции VLOOKUP в Python

В нашем наборе данных, на одном из листов Excel, есть названия городов, а на другом — названия штатов и провинций. Как узнать о том, где именно находится каждый город? Для этого подходит Excel-функция VLOOKUP, с помощью которой можно связать данные двух таблиц. Эта функция работает по принципу левого соединения, когда сохраняется каждая запись из набора данных, находящегося в левой части выражения. Применяя функцию VLOOKUP, мы предлагаем системе выполнить поиск определённого значения в заданном столбце указанного листа, а затем — вернуть значение, которое находится на заданное число столбцов правее найденного значения. Вот как это выглядит:

=VLOOKUP([@City],states,2,false)

Зададим на листе sales заголовок столбца F как State и воспользуемся функцией VLOOKUP для того чтобы заполнить ячейки этого столбца названиями штатов и провинций, в которых расположены города.

Использование функции VLOOKUP в Excel

В Python сделать то же самое можно, воспользовавшись методом merge из pandas. Он принимает два датафрейма и объединяет их. Для решения этой задачи нам понадобится следующий код:

sales = pd.merge(sales, states, how='left', on='City')

Разберём его:

1. Первый аргумент метода merge — это исходный датафрейм.
2. Второй аргумент — это датафрейм, в котором мы ищем значения.
3. Аргумент how указывает на то, как именно мы хотим соединить данные.
4. Аргумент on указывает на переменную, по которой нужно выполнить соединение (тут ещё можно использовать аргументы left_on и right_on, нужные в том случае, если интересующие нас данные в разных датафреймах названы по-разному).

Сводные таблицы

Сводные таблицы (Pivot Tables) — это одна из самых мощных возможностей Excel. Такие таблицы позволяют очень быстро извлекать ценные сведения из больших наборов данных. Создадим в Excel сводную таблицу, выводящую сведения о суммарных продажах по каждому городу.

Создание сводной таблицы в Excel

Как видите, для создания подобной таблицы достаточно перетащить поле City в раздел Rows, а поле Sales — в раздел Values. После этого Excel автоматически выведет суммарные продажи для каждого города.

Для того чтобы создать такую же сводную таблицу в pandas, нужно будет написать следующий код:

sales.pivot_table(index = 'City', values = 'Sales', aggfunc = 'sum')

Разберём его:

1. Здесь мы используем метод sales.pivot_table, сообщая pandas о том, что мы хотим создать сводную таблицу, основанную на датафрейме sales.
2. Аргумент index указывает на столбец, по которому мы хотим агрегировать данные.
3. Аргумент values указывает на то, какие значения мы собираемся агрегировать.
4. Аргумент aggfunc задаёт функцию, которую мы хотим использовать при обработке значений (тут ещё можно воспользоваться функциями mean, max, min и так далее).

Итоги

Из этого материала вы узнали о том, как импортировать Excel-данные в pandas, о том, как реализовать средствами Python и pandas возможности Excel-функций IF и VLOOKUP, а также о том, как воспроизвести средствами pandas функционал сводных таблиц Excel. Возможно, сейчас вы задаётесь вопросом о том, зачем вам пользоваться pandas, если то же самое можно сделать и в Excel. На этот вопрос нет однозначного ответа. Python позволяет создавать код, который поддаётся тонкой настройке и глубокому исследованию. Такой код можно использовать многократно. Средствами Python можно описывать очень сложные схемы анализа данных. А возможностей Excel, вероятно, достаточно лишь для менее масштабных исследований данных. Если вы до этого момента пользовались только Excel — рекомендую испытать Python и pandas, и узнать о том, что у вас из этого получится.

А какие инструменты вы используете для анализа данных?

Напоминаем, что у нас продолжается конкурс прогнозов, в котором можно выиграть новенький iPhone. Еще есть время ворваться в него, и сделать максимально точный прогноз по злободневным величинам.

Last updated on 
Jul 18, 2021

In this post you can learn how to read Excel files (ext xls, xlsx etc) with Python and Pandas. We will import one or several sheets from an Excel file to a Pandas DataFrame.

The list of the supported file extensions:

• xls
• xlsx
• xlsm
• xlsb
• odf
• ods
• odt

Note for ods, ods and odt please check: Read Excel(OpenDocument ODS) with Python Pandas

Step 1: Install Pandas and odfpy

Python offers many different modules for reading and manipulating Excel files. In this guide we are going to use pandas and odfpy:

pip install pandas
pip install odfpy

Step 2: Read the one sheet of Excel(XLS) file

Pandas offers a powerful method for reading any type of Excel files read_excel(). It’s pretty easy to be used and requires only the file path:

import pandas as pd

pd.read_excel('animals.xls')

It will read and return all non empty cells from the Excel file:

	Rank	Animal	Maximum speed	Class	Notes
0	1	Peregrine falcon	389 km/h (242 mph)108 m/s (354 ft/s)[2][6]	Flight-diving	The peregrine falcon is the fastest aerial ani…
1	2	Golden eagle	240–320 km/h (150–200 mph)67–89 m/s (220–293 f…	Flight-diving	Assuming the maximum size at 1.02 m, its relat…
2	3	White-throated needletail swift	169 km/h (105 mph)[8][9][10]	Flight	NaN
3	4	Eurasian hobby	160 km/h (100 mph)[11]	Flight	Can sometimes outfly the swift
4	5	Mexican free-tailed bat	160 km/h (100 mph)[12]	Flight	It has been claimed to have the fastest horizo…
5	6	Frigatebird	153 km/h (95 mph)	Flight	The frigatebird’s high speed is helped by its …
6	7	Rock dove (pigeon)	148.9 km/h (92.5 mph)[13]	Flight	Pigeons have been clocked flying 92.5 mph (148…
7	8	Spur-winged goose	142 km/h (88 mph)[14]	Flight	NaN
8	9	Gyrfalcon	128 km/h (80 mph)[citation needed]	Flight	NaN

Step 3: Read the second sheet of Excel file by name

If you like to read data from a specific sheet — for example Sheet 2 then you can specify the name as a parameter — sheet_name:

pd.read_excel('animals.xlsx', sheet_name="Sheet2")

Which will result in:

	Blackbuck	Unnamed: 1
0	NaN	NaN
1	Male blackbuck	Male blackbuck
2	NaN	NaN
3	Female with young at the National Zoological Park Delhi	Female with young at the National Zoological P…
4	Conservation status	Conservation status
5	Least Concern (IUCN 3.1)[1]	Least Concern (IUCN 3.1)[1]
6	Scientific classification	Scientific classification

Step 4: Python read excel file — specify columns and rows

If you like to read a range of data and not the whole sheet — read_excel offers several very useful parameters.

Python read excel file select rows

Next code example will show you how to read 3 rows skipping the first two rows. In this way Pandas will read only some rows from the whole sheet:

pd.read_excel('animals.xlsx', skiprows=2, nrows=3)

which will result in:

	2	Golden eagle	240–320 km/h (150–200 mph)67–89 m/s (220–293 f…	Flight-diving	Assuming the maximum size at 1.02 m, its relat…
0	3	White-throated needletail swift	169 km/h (105 mph)[8][9][10]	Flight	NaN
1	4	Eurasian hobby	160 km/h (100 mph)[11]	Flight	Can sometimes outfly the swift
2	5	Mexican free-tailed bat	160 km/h (100 mph)[12]	Flight	It has been claimed to have the fastest horizo…

Python read excel file select columns

If you like to** work with few columns** and not the whole sheet — then parameter use_cols can be used as shown:

pd.read_excel('animals.xlsx', usecols='C:D')

Python read excel file specify columns and rows

Finally if you like to select a range from specific columns and rows than you can use:

Which will result into:

	240–320 km/h (150–200 mph)67–89 m/s (220–293 f…	Flight-diving
0	169 km/h (105 mph)[8][9][10]	Flight
1	160 km/h (100 mph)[11]	Flight
2	160 km/h (100 mph)[12]	Flight

Step 5. Read multiple sheets from Excel file

What if you like to read with Pandas multiple sheets from Excel. It’s possible with pd.read_excel by providing a list of all sheets to be read as follows:

pd.read_excel('animals.xlsx', sheet_name=["Sheet1", "Sheet2"])

Note that a dictionary of

• keys — sheet names
• values — resulted DataFrames

will be returned.

In order to access data you can access it by a sheet name as:

pd.read_excel('animals.xlsx', sheet_name=["Sheet1", "Sheet2"]).get('Sheet1')

which will return the data for Sheet1 as a DataFrame.

Read All Sheets

For loading all sheets from Excel file use sheet_name=None:

pd.read_excel('animals.xlsx', sheet_name=None)

Step 6. Pandas read excel data with conversion, NA values and parsing

Finally let’s check what we can do if we need to convert data, drop or fill missing values, parse dates and numbers.

Pandas offers several parameters for this purpose:

• converters — dict of functions for converting values in certain columns
• keep_default_na — whether or not to include the default NaN values
• parse_dates
• ate_parser — converting a sequence of string columns to an array of datetime instances.
• thousands
• convert_float

You can check the Notebook in the resources for more examples of the above.

Resources

• Python Pandas Reading Excel files
• pandas.read_excel
• Notebook —
  Read Excel ODS with Python Pandas