Basic usage of pandas
This article introduces the relevant knowledge of "the basic usage of pandas". Many people will encounter such a dilemma in the operation of actual cases, so let the editor lead you to learn how to deal with these situations. I hope you can read it carefully and be able to achieve something!
Create pandas1. Use the pd.Series () function to create DataFrameimport pandas as pdcountry1 = pd.Series ({'Name':' China', 'Language':' Chinese', 'Area':' 9.597M km2', 'Happiness Rank': 79}) country2 = pd.Series ({' Name': 'USA') 'Language':' English (US)', 'Area':' 9.834M km2', 'Happiness Rank': 14}) country3 = pd.Series ({' Name': 'Australia', 'Language':' English (AU)', 'Area':' 7.692m km2' 'Happiness Rank': 9}) df = pd.DataFrame ([country1, country2, country3], index= [' CH', 'US',' AU'])
two。 Use DataFrame () and dictionary to create df = pd.DataFrame (columns= ["epoch", "train_loss", "train_auc", "test_loss", "test_auc"]) log_dic = {"epoch": 1, "train_loss": 0.2, "train_auc": 1, "test_loss": 0 "test_auc": 0} df = df.append ([log_dic]) log_dic = {"epoch": 2, "train_loss": 0.2, "train_auc": 1, "test_loss": 0 "test_auc": 0} df = df.append ([log_dic]) # renumber index # inplace=True means to modify # drop=True on the original data to indicate the indexdf.reset_index (inplace=True, drop=True) before discarding
3. Using pd.concat to combine multiple DataFrame () df1 = pd.DataFrame (columns= ["epoch", "train_loss", "train_auc", "test_loss", "test_auc"]) log_dic = {"epoch": 1, "train_loss": 0.2, "train_auc": 1, "test_loss": 0 "test_auc": 0} df1 = df1.append ([log_dic]) df2 = pd.DataFrame (columns= ["epoch", "train_loss", "train_auc", "test_loss", "test_auc"]) log_dic = {"epoch": 2, "train_loss": 0.1, "train_auc": 1, "test_loss": 0 "test_auc": 1} df2 = df2.append ([log_dic]) # ignore_index=True means renumber index df_new = pd.concat ([df1, df2], axis=0, ignore_index=True)
File operation 1. Generate a file columns = ["epoch", "train_loss", "train_auc", "test_loss", "test_auc"] df_ [header] .to _ csv ('text.txt', index=False, header=columns, sep='\ t') with columns
two。 Generate a file df_ new [header] .to _ csv ('text.txt', index=False, header=None, sep='\ t') that does not contain columns
3. Read the file df = pd.read_csv ('text.txt', sep='\ tasking, header=None, nrows=100) df.columns = ["epoch", "train_loss", "train_auc", "test_loss", "test_auc"]
4. To read a file containing columns # you need to use the header parameter to specify the line in which the columns is on, usually line 0 df = pd.read_csv ('text.txt', sep='\ tasking, header= [0]) # specify a specific columns read reprot_2016_df = pd.read_csv (' 2016.csvault, index_col='Country', usecols= ['Country',' Happiness Rank' 'Happiness Score',' Region'])
5. Generate a binary pickle file df = pd.DataFrame (columns= ["epoch", "train_loss", "train_auc", "test_loss", "test_auc"]) log_dic = {"epoch": 2, "train_loss": 0.1, "train_auc": 1, "test_loss": 23 "test_auc": 1} df = df.append ([log_dic]) df.to_pickle ('df_log.pickle')
6. Load pickle Fil
Df = pd.read_pickle ('df_log.pickle') index
Use the data in the following figure as an example
1. Row index df.loc ['CH'] # Series type
Df.loc ['CH']. Index # Index ([' Name', 'Language',' Area', 'Happiness Rank'], dtype='object') df.loc [' CH'] ['Name'] #' China 'df.loc [' CH']. To_numpy () # array (['China', 'Chinese',' 9.597M km2', 79], dtype=object) df.iloc [1] # second row of the index
Df.loc [['CH',' US']] df.iloc [[0,1]]
two。 Column index df ['Area'] # type: Seriesdf [[' Name', 'Area']] # type: DataFrame
3. Mixed index print ('fetch column first, then row:') print (df ['Area'] [' CH']) print (df ['Area'] .loc [' CH']) print (df ['Area'] .iloc [0]) print (' fetch row first, then fetch column:') print (df.loc ['CH'] [' Area']) print (df.iloc [0] ['Area']) print (df.at [' CH', 'Area'])
Delete data df.drop (['CH'], inplace=True) # Delete row inplace=True means modify df.drop ([' Area'], axis=1, inplace=True) # delete column, you need to specify operation 1 of axis=1Nan data. Delete operation
Use the following data
Import numpy as npdf = pd.DataFrame ({"name": ['Alfred',' Batman', 'Catwoman'], "toy": [np.nan,' Batmobile', 'Bullwhip'], "born": [pd.NaT, pd.Timestamp ("1940-04-25"), pd.NaT]})
"" axis: 0: row operation (default) 1: column operation how: any: delete whenever there is a null value (default) all: delete inplace: False: return a new dataset (default) True: operate on the desired dataset "" df.dropna (axis=0, how='any', inplace=True)
Df.dropna (axis=0, how='any', subset= ['toy'], inplace=False) # subset specifies the nan2 that operates on a specific column. Filling operation
Use the following data
Df = pd.DataFrame ([[np.nan, 2, np.nan, 0], [3,4, np.nan, 1], [np.nan, 5], [np.nan, 3, np.nan, 4]], columns=list ('ABCD')) df.fillna (0, inplace=True)
# "replace the missing value horizontally with the value preceding the missing value" df.fillna (axis=1, method='ffill', inplace=False)
# "replace the missing value vertically with the value above the missing value" df.fillna (axis=0, method='bfill', inplace=False)
Df ['A'] .fillna (0, inplace=True) # specifies a specific column fill
3. Judge operation df.isnull () df ['A'] .isna ()
Merger operation import pandas as pdstaff_df = pd.DataFrame ({'name': 'Zhang San', 'department': 'research and development department'}, {'name':'Li Si', 'department': 'finance department'}, {'name': 'Zhao Liu' 'Department': 'Marketing Department'}]) student_df = pd.DataFrame ([{'name': 'Zhang San', 'major': 'computer'}, {'name':'Li Si', 'major': 'accountant'}, {'name': 'Wang Wu', 'specialty:' marketing'}]) 1. Merge according to column benchmark
Inner (intersection) outer (union) left right
Pd.merge (staff_df, student_df, how='inner', on=' name') pd.merge (staff_df, student_df, how='outer', on=' name')
two。 Merge # to index staff_df.set_index ('name', inplace=True) student_df.set_index ('name', inplace=True) pd.merge (staff_df, student_df, how='left', left_index=True, right_index=True)
3. Specify a different column benchmark # reset index to range () staff_df.reset_index (inplace=True) student_df.reset_index (inplace=True) staff_df.rename (columns= {'name': 'employee name'}, inplace=True) student_df.rename (columns= {'name': 'student name'}, inplace=True) pd.merge (staff_df, student_df, how='left', left_on=' employee name', right_on=' student name') pd.merge (staff_df Student_df, how='inner', left_on= ['employee name', 'address'], right_on= ['student name', 'address']) other operations report_data = pd.read_csv ('. / 2015.csv') report_data.head () basic attribute operation data.head () data.info () data.describe () data.columnsdata.index rename columnsdf.rename (columns= {'Region':' region', 'Happiness Rank':' ranking' 'Happiness Score':' Happiness Index'}, inplace=True) data cleaning operation # null is replaced with 0df.fillna (0 Inplace=False) # discard nulldf.dropna () # forward fill df.ffill () # backward fill df.bfill (inplace=True) apply operation # apply use # get last name staff_df ['employee name'] .apply (lambda x: X [0]) # get the name staff_df ['employee name'] .apply (lambda x: X [1:]) # result merge staff_df.loc [: 'Last name'] = staff_df ['employee name'] .apply (lambda x: X [0]) staff_df.loc [:, 'first name'] = staff_df ['employee name'] .apply (lambda x: X [1:]) grouping
Grouping according to columns
Grouped = report_data.groupby ('Region') grouped [' Happiness Score'] .mean () # iterative groupby object for group, frame in grouped: mean_score = frame ['Happiness Score'] .mean () max_score = frame [' Happiness Score'] .max () min_score = frame ['Happiness Score'] .min () print (average happiness index of' {} region: {}, highest happiness index: {} Minimum happiness index {} '.format (group, mean_score, max_score, min_score)
Define function grouping
Report_data2 = report_data.set_index ('Happiness Rank') def get_rank_group (rank): rank_group =''if rank 2'if the custom function The operation is aimed at indexreport_data ['score group'] = report_data [' Happiness Score'] .apply (lambda score: int (score)) grouped = report_data.groupby ('score group') for group, frame in grouped: print ({}' .format (group, len (frame)
Use a bar chart of type bar to count the number of each label.
This is the end of train_df.label.value_counts (). Plot (kind='bar') "basic usage of pandas". Thank you for reading. If you want to know more about the industry, you can follow the website, the editor will output more high-quality practical articles for you!
