Missing Values - Weather Data:

-----------------------------------------------------------------------------------
--------------------------------------------------
# Importing the libraries

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

# Ignore harmless warnings

import warnings
warnings.filterwarnings("ignore")

# Set to display all the columns in dataset

pd.set_option("display.max_columns", None)

# Import psql to run queries

import pandasql as psql

-----------------------------------------------------------------------------------
--------------------------------------------------
# Load the weather dataset

weather = pd.read_csv(r"E:\R3SPAnalytics\00 IIT KGP Hyd\00-LATA\Sessions\Session-

07\weather.csv", header=0)

# Copy to back-up file

weather_bk = weather.copy()

# Display first 5 records

weather.head()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display the dataset information

weather.info()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Identify the missing data in all variables

weather.isnull().sum()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Use KNNImputer to address missing values

from sklearn.impute import KNNImputer

imputer_int = KNNImputer(missing_values=np.nan, n_neighbors=5, weights='uniform',

metric='nan_euclidean',
copy=True, add_indicator=False)

weather['Sunshine'] = imputer_int.fit_transform(weather[['Sunshine']])
-----------------------------------------------------------------------------------
--------------------------------------------------
# Use SimpleImputer to address missing values

from sklearn.impute import SimpleImputer

imputer_str = SimpleImputer(missing_values=np.nan, strategy='most_frequent',

fill_value=None, verbose=0,
copy=True, add_indicator=False)
#Claims_BK1['Policy_Type'] =
imputer_si.fit_transform(Claims_BK1['Policy_Type'].values.reshape(-1,1))[:,0]
weather['WindGustDir'] =
imputer_str.fit_transform(weather['WindGustDir'].values.reshape(-1,1))[:,0]
weather['WindGustSpeed'] =
imputer_int.fit_transform(weather['WindGustSpeed'].values.reshape(-1,1))[:,0]
weather['WindDir9am'] =
imputer_str.fit_transform(weather['WindDir9am'].values.reshape(-1,1))[:,0]
weather['WindDir3pm'] =
imputer_str.fit_transform(weather['WindDir3pm'].values.reshape(-1,1))[:,0]
weather['WindSpeed9am'] =
imputer_int.fit_transform(weather['WindSpeed9am'].values.reshape(-1,1))[:,0]
-----------------------------------------------------------------------------------
--------------------------------------------------
# Drop the RISK_MM variable

weather.drop('RISK_MM', inplace=True, axis=1)

weather.head()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Info after addressing the missing values

weather.info()
-----------------------------------------------------------------------------------
--------------------------------------------------
# "Groupby" we have important information about the possible conditions that may or
may not cause rain condition

weather[['Rainfall','Sunshine','Evaporation','WindGustSpeed','WindSpeed9am','WindSp
eed3pm','Humidity9am',
'Humidity3pm','Pressure9am','Pressure3pm','Cloud9am','Cloud3pm','Temp9am',
'Temp3pm']].groupby(weather['RainToday']).mean()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display the WindGustDir by sub-variable count

weather['WindGustDir'].value_counts()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Check the null values

weather.isnull().sum()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display the 'weather' dataset variables by sub-variable count

for i in weather.columns:
print(weather[i].value_counts())
-----------------------------------------------------------------------------------
--------------------------------------------------
# Identify the numerical and categorical variables
num_vars = weather.columns[weather.dtypes != 'object']
cat_vars = weather.columns[weather.dtypes == 'object']
print(num_vars)
print(cat_vars)
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display the WindGustDir by sub-variable count

weather['WindGustDir'].value_counts()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Use LabelEncoder to handle categorical data

from sklearn.preprocessing import LabelEncoder

LE = LabelEncoder()

weather['WindGustDir'] = LE.fit_transform(weather[['WindGustDir']])
weather['WindDir9am'] = LE.fit_transform(weather[['WindDir9am']])
weather['WindDir3pm'] = LE.fit_transform(weather[['WindDir3pm']])
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display the RainToday by sub-variable count

weather['RainToday'].value_counts()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display the RainTomorrow by sub-variable count

weather['RainTomorrow'].value_counts()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Use LabelBinarizer for target variables

from sklearn.preprocessing import LabelBinarizer

LB = LabelBinarizer()

weather['RainToday'] = LB.fit_transform(weather['RainToday'])
weather['RainTomorrow'] = LB.fit_transform(weather['RainTomorrow'])

weather.head()
-----------------------------------------------------------------------------------
--------------------------------------------------

-----------------------------------------------------------------------------------
--------------------------------------------------

-----------------------------------------------------------------------------------
--------------------------------------------------

-----------------------------------------------------------------------------------
--------------------------------------------------

-----------------------------------------------------------------------------------
--------------------------------------------------

-----------------------------------------------------------------------------------
--------------------------------------------------
-----------------------------------------------------------------------------------
--------------------------------------------------

-----------------------------------------------------------------------------------
--------------------------------------------------

-----------------------------------------------------------------------------------
--------------------------------------------------