Key Error: None of [Int64Index...] dtype='int64] are in the columns

Question

I'm trying to shuffle my indices using the np.random.shuffle() method, but I keep getting an error that I don't understand. I'd appreciate it if someone could help me puzzle this out. Thank you!

I've tried to use the delimiter=',' and delim_whitespace=0 when I made my raw_csv_data variable at the beginning, as I saw that as the solution of another problem, but it kept throwing the same error

    import pandas as pd 
    import numpy as np 
    from sklearn.preprocessing import StandardScaler

    #%%
    raw_csv_data= pd.read_csv('Absenteeism-data.csv')
    print(raw_csv_data)
    #%%
    df= raw_csv_data.copy()
    print(display(df))
    #%%
    pd.options.display.max_columns=None
    pd.options.display.max_rows=None
    print(display(df))
    #%%
    print(df.info())
    #%%
    df=df.drop(['ID'], axis=1)

    #%%
    print(display(df.head()))

    #%%
    #Our goal is to see who is more likely to be absent. Let's define
    #our targets from our dependent variable, Absenteeism Time in Hours
    print(df['Absenteeism Time in Hours'])
    print(df['Absenteeism Time in Hours'].median())
    #%%
    targets= np.where(df['Absenteeism Time in Hours']>df['Absenteeism Time 
    in Hours'].median(),1,0)
    #%%
    print(targets)
    #%%
    df['Excessive Absenteeism']= targets
    #%%
    print(df.head())

    #%%
    #Let's Separate the Day and Month Values to see if there is 
    correlation
    #between Day of week/month with absence
    print(type(df['Date'][0]))
    #%%
    df['Date']= pd.to_datetime(df['Date'], format='%d/%m/%Y')
    #%%
    print(df['Date'])
    print(type(df['Date'][0]))
    #%%
    #Extracting the Month Value
    print(df['Date'][0].month)
    #%%
    list_months=[]
    print(list_months)
    #%%
    print(df.shape)
    #%%
    for i in range(df.shape[0]):
        list_months.append(df['Date'][i].month)
    #%%
    print(list_months)
    #%%
    print(len(list_months))
    #%%
    #Let's Create a Month Value Column for df
    df['Month Value']= list_months
    #%%
    print(df.head())
    #%%
    #Now let's extract the day of the week from date
    df['Date'][699].weekday()
    #%%
    def date_to_weekday(date_value):
        return date_value.weekday()
    #%%
    df['Day of the Week']= df['Date'].apply(date_to_weekday)
    #%%
    print(df.head())
    #%%
    df= df.drop(['Date'], axis=1)
    #%%
    print(df.columns.values)
    #%%
    reordered_columns= ['Reason for Absence', 'Month Value','Day of the 
    Week','Transportation Expense', 'Distance to Work', 'Age',
     'Daily Work Load Average', 'Body Mass Index', 'Education', 
    'Children', 
    'Pets',
     'Absenteeism Time in Hours', 'Excessive Absenteeism']
    #%%
    df=df[reordered_columns]
    print(df.head())
    #%%
    #First Checkpoint
    df_date_mod= df.copy()
    #%%
    print(df_date_mod)

    #%%
    #Let's Standardize our inputs, ignoring the Reasons and Education 
    Columns
    #Because they are labelled by a separate categorical criteria, not 
    numerically
    print(df_date_mod.columns.values)
    #%%
    unscaled_inputs= df_date_mod.loc[:, ['Month Value','Day of the 
    Week','Transportation Expense','Distance to Work','Age','Daily Work 
    Load 
    Average','Body Mass Index','Children','Pets','Absenteeism Time in 
    Hours']]
    #%%
    print(display(unscaled_inputs))
    #%%
    absenteeism_scaler= StandardScaler()
    #%%
    absenteeism_scaler.fit(unscaled_inputs)
    #%%
    scaled_inputs= absenteeism_scaler.transform(unscaled_inputs)
    #%%
    print(display(scaled_inputs))
    #%%
    print(scaled_inputs.shape)
    #%%
    scaled_inputs= pd.DataFrame(scaled_inputs, columns=['Month Value','Day 
    of the Week','Transportation Expense','Distance to Work','Age','Daily 
    Work Load Average','Body Mass Index','Children','Pets','Absenteeism 
    Time 
    in Hours'])
    print(display(scaled_inputs))
    #%%
    df_date_mod= df_date_mod.drop(['Month Value','Day of the 
    Week','Transportation Expense','Distance to Work','Age','Daily Work 
    Load Average','Body Mass Index','Children','Pets','Absenteeism Time in 
    Hours'], axis=1)
    print(display(df_date_mod))
    #%%
    df_date_mod=pd.concat([df_date_mod,scaled_inputs], axis=1)
    print(display(df_date_mod))
    #%%
    df_date_mod= df_date_mod[reordered_columns]
    print(display(df_date_mod.head()))
    #%%
    #Checkpoint
    df_date_scale_mod= df_date_mod.copy()
    print(display(df_date_scale_mod.head()))
    #%%
    #Let's Analyze the Reason for Absence Category
    print(df_date_scale_mod['Reason for Absence'])
    #%%
    print(df_date_scale_mod['Reason for Absence'].min())
    print(df_date_scale_mod['Reason for Absence'].max())
    #%%
    print(df_date_scale_mod['Reason for Absence'].unique())
    #%%
    print(len(df_date_scale_mod['Reason for Absence'].unique()))
    #%%
    print(sorted(df['Reason for Absence'].unique()))
    #%%
    reason_columns= pd.get_dummies(df['Reason for Absence'])
    print(reason_columns)
    #%%
    reason_columns['check']= reason_columns.sum(axis=1)
    print(reason_columns)
    #%%
    print(reason_columns['check'].sum(axis=0))
    #%%
    print(reason_columns['check'].unique())
    #%%
    reason_columns=reason_columns.drop(['check'], axis=1)
    print(reason_columns)
    #%%
    reason_columns=pd.get_dummies(df_date_scale_mod['Reason for Absence'], 
    drop_first=True)
    print(reason_columns)
    #%%
    print(df_date_scale_mod.columns.values)
    #%%
    print(reason_columns.columns.values)
    #%%
    df_date_scale_mod= df_date_scale_mod.drop(['Reason for Absence'], 
    axis=1)
    print(df_date_scale_mod)
    #%%
    reason_type_1= reason_columns.loc[:, 1:14].max(axis=1)
    reason_type_2= reason_columns.loc[:, 15:17].max(axis=1)
    reason_type_3= reason_columns.loc[:, 18:21].max(axis=1)
    reason_type_4= reason_columns.loc[:, 22:].max(axis=1)
    #%%
    print(reason_type_1)
    print(reason_type_2)
    print(reason_type_3)
    print(reason_type_4)
    #%%
    print(df_date_scale_mod.head())
    #%%
    df_date_scale_mod= pd.concat([df_date_scale_mod, 
    reason_type_1,reason_type_2, reason_type_3, reason_type_4], axis=1)
    print(df_date_scale_mod.head())
    #%%
    print(df_date_scale_mod.columns.values)
    #%%
    column_names= ['Month Value','Day of the Week','Transportation 
    Expense',
     'Distance to Work','Age','Daily Work Load Average','Body Mass Index',
     'Education','Children','Pets','Absenteeism Time in Hours',
     'Excessive Absenteeism', 'Reason_1', 'Reason_2', 'Reason_3', 
     'Reason_4']

    df_date_scale_mod.columns= column_names
    print(df_date_scale_mod.head())
    #%%
    column_names_reordered= ['Reason_1', 'Reason_2', 'Reason_3', 
    'Reason_4','Month Value','Day of the Week','Transportation Expense',
     'Distance to Work','Age','Daily Work Load Average','Body Mass Index',
     'Education','Children','Pets','Absenteeism Time in Hours',
     'Excessive Absenteeism']

    df_date_scale_mod=df_date_scale_mod[column_names_reordered]
    print(display(df_date_scale_mod.head()))
    #%%
    #Checkpoint
    df_date_scale_mod_reas= df_date_scale_mod.copy()
    print(df_date_scale_mod_reas.head())
    #%%
    #Let's Look at the Education column now
    print(df_date_scale_mod_reas['Education'].unique())
    #This shows us that education is rated from 1-4 based on level
    #of completion
    #%%
    print(df_date_scale_mod_reas['Education'].value_counts())
    #The overwhelming majority of workers are highschool educated, while 
    the 
    #rest have higher degrees
    #%%
    #We'll create our dummy variables as highschool and higher education
    df_date_scale_mod_reas['Education']= 
    df_date_scale_mod_reas['Education'].map({1:0, 2:1, 3:1, 4:1})
    #%%
    print(df_date_scale_mod_reas['Education'].unique())
    #%%
    print(df_date_scale_mod_reas['Education'].value_counts())
    #%%
    #Checkpoint
    df_preprocessed= df_date_scale_mod_reas.copy()
    print(display(df_preprocessed.head()))
    #%%
    #%%
    #Split Inputs from targets
    scaled_inputs_all= df_preprocessed.loc[:,'Reason_1':'Absenteeism Time 
    in 
    Hours']
    print(display(scaled_inputs_all.head()))
    print(scaled_inputs_all.shape)
    #%%
    targets_all= df_preprocessed.loc[:,'Excessive Absenteeism']
    print(display(targets_all.head()))
    print(targets_all.shape)
    #%%
    #Shuffle Inputs and targets
    shuffled_indices= np.arange(scaled_inputs_all.shape[0])
    np.random.shuffle(shuffled_indices)
    shuffled_inputs= scaled_inputs_all[shuffled_indices]
    shuffled_targets= targets_all[shuffled_indices]

This is the error I keep getting when I try to shuffle my indices:

KeyError                                  Traceback (most recent call last)
 in 
      1 shuffled_indices= np.arange(scaled_inputs_all.shape[0])
      2 np.random.shuffle(shuffled_indices)
----> 3 shuffled_inputs= scaled_inputs_all[shuffled_indices]
      4 shuffled_targets= targets_all[shuffled_indices]

~\Anaconda3\lib\site-packages\pandas\core\frame.py in getitem(self, key) 2932 key = list(key) 2933 indexer = self.loc._convert_to_indexer(key, axis=1, -> 2934 raise_missing=True) 2935 2936 # take() does not accept boolean indexers

~\Anaconda3\lib\site-packages\pandas\core\indexing.py in _convert_to_indexer(self, obj, axis, is_setter, raise_missing) 1352 kwargs = {'raise_missing': True if is_setter else 1353
raise_missing} -> 1354 return self._get_listlike_indexer(obj, axis, **kwargs)[1] 1355 else: 1356 try:

~\Anaconda3\lib\site-packages\pandas\core\indexing.py in _get_listlike_indexer(self, key, axis, raise_missing) 1159 self._validate_read_indexer(keyarr, indexer, 1160
o._get_axis_number(axis), -> 1161 raise_missing=raise_missing) 1162 return keyarr, indexer
1163

~\Anaconda3\lib\site-packages\pandas\core\indexing.py in _validate_read_indexer(self, key, indexer, axis, raise_missing) 1244 raise KeyError( 1245
u"None of [{key}] are in the [{axis}]".format( -> 1246 key=key, axis=self.obj._get_axis_name(axis))) 1247 1248 # We (temporarily) allow for some missing keys with .loc, except in

KeyError: "None of [Int64Index([560, 320, 405, 141, 154, 370, 656, 26, 444, 307,\n ...\n 429, 542, 676, 588, 315, 284, 293, 607, 197, 250],\n dtype='int64', length=700)] are in the [columns]"

Solution 1

You created your scaled_inputs_all DataFrame using loc function, so it most likely contains no consecutive indices.

On the other hand, you created shuffled_indices as a shuffle from just a range of consecutive numbers.

Remember that scaled_inputs_all[shuffled_indices] gets rows of scaled_inputs_all which have index values equal to elements of shuffled_indices.

Maybe you should write:

scaled_inputs_all.iloc[shuffled_indices]

Note that iloc provides integer-location based indexing, regardless of index values, i.e. just what you need.

Solution 2

might have someone also get the same error in working with KFOLD in machine learning.

And the solution for this is as below:

Click here to watch solutinon

You need to use iloc:

 X_train, X_test = X.iloc[train_index], X.iloc[test_index]

 y_train, y_test = y.iloc[train_index], y.iloc[test_index]

Solution 3

I had this problem too. I solved it by changing the data frame and series to array.

try the follwing codeline:

scaled_inputs_all.iloc[shuffled_indices].values 

Solution 4

If you reset your index after dropping rows from your dataframe, this should stop the key error.

You can do this by running this after you run df.drop:

df = df.reset_index(drop=True)

Or, equivalently:

df.reset_index(drop=True, inplace=True)

View more solutions

Comments

[Nilesh Ingle]

A reproducible example with few rows of dataframe would help to look into this issue.

[Ashley E.]

@NileshIngle do you want me to send you the dataset that I was working with?

[Ashley E.]

omg it worked! Thank you! Can you please explain why again? I didn't really understand your explanation up above

[Ashley E.]

would I need to write targets_all.iloc[shuffled_indices] too?

[Valdi_Bo]

Yes, but make sure that both DataFrames contain the same number of rows. Otherwise you may get index out of range error. Maybe the range of indices should be different.

[mayaaa]

i have the same error when i run the following code. could you tell me how to solve it skf = StratifiedKFold(n_splits=10, shuffle=True, random_state=0) for train_index, test_index in skf.split(x, y): x_train, x_test = x[train_index], x[test_index] y_train, y_test = y[train_index], y[test_index]

[Sean William]

This one solved my problem!