vlambda博客
学习文章列表

#Import Library and Load Fileimport pandas as pdimport numpy as npdf = pd.read_csv('/kaggle/input/mall-customers/Mall_Customers.csv')df.info() #checking data types and total null values

#Checking If any duplicated valuesprint(f'Total Duplicated Rows : {df.duplicated().sum()}')

#Let's see the percentile from each numerical columns from the dataset
def percentile(df, column) : print(f'{column} Percentile Summary :') for a in range(0,101,10) : print(f'- {a}th Percentile : {round(np.percentile(df[column],a),2)}') #Percentile for Agepercentile(df, 'Age')#Annual Income Percentilepercentile(df,'Annual Income (k$)')#Spending Score Percentilepercentile(df,'Spending Score (1-100)')

#Count Each Gender totalgender_total = df['Genre'].value_counts().reset_index()gender_total['perc_genre'] = round(gender_total['Genre']/sum(gender_total['Genre']),2)*100gender_total

import matplotlib.pyplot as pltimport seaborn as snsimport plotly.express as px
num_cols = ['Age','Annual Income (k$)','Spending Score (1-100)']def plot_stats(df, col_list) :for a in num_cols :fig,ax = plt.subplots(1,2, figsize = (9,6))
sns.distplot(df[a], ax = ax[0])sns.boxplot(df[a], ax = ax[1])
ax[0].axvline(df[a].mean(), linestyle = '--', linewidth = 2, color = 'green')ax[0].axvline(df[a].median(), linestyle = '--', linewidth = 2 , color = 'red')
ax[0].set_ylabel('Frequency')ax[0].set_title('Distribution Plot')
ax[1].set_title('Box Plot')
plt.suptitle(a)plt.show()
plot_stats(df, num_cols)

#Flooring and Capping by replacing outliers with 10th and 90th Percentile
#Age 10th Percentile and 90th Percentiletenth_percentile_age = np.percentile(df['Age'], 10)ninetieth_percentile_age = np.percentile(df['Age'], 90)
df['Age'] = np.where(df['Age'] < tenth_percentile_age, tenth_percentile_age, df['Age'])df['Age'] = np.where(df['Age'] > ninetieth_percentile_age, ninetieth_percentile_age, df['Age'])
#Annual Income 10th Percentile and 90th Percentiletenth_percentile_annualincome = np.percentile(df['Annual Income (k$)'], 10)ninetieth_percentile_annualincome = np.percentile(df['Annual Income (k$)'], 90)df['Annual Income (k$)'] = np.where(df['Annual Income (k$)'] < tenth_percentile_annualincome, tenth_percentile_annualincome, df['Annual Income (k$)'])df['Annual Income (k$)'] = np.where(df['Annual Income (k$)'] > ninetieth_percentile_annualincome, ninetieth_percentile_annualincome, df['Annual Income (k$)'])plot_stats(df, num_cols) #Checking Distribution after replacing outliers with 10th and 90th Percentile

from sklearn.preprocessing import MinMaxScalerfrom sklearn.decomposition import PCAfrom sklearn.cluster import KMeans#Normalize Numeric Featuresscaled_features = MinMaxScaler().fit_transform(df.iloc[:,3:5])
#Get 2 Principal Componentspca = PCA(n_components = 2).fit(scaled_features)features_2d = pca.transform(scaled_features)#5 Centroids Modelmodel = KMeans(n_clusters = 5, init= 'k-means++', n_init = 100, max_iter = 1000, random_state=16)
#Fit to the data and predict the cluster assignments to each data pointsfeature = df.iloc[:,3:5]km_clusters = model.fit_predict(feature.values)km_clusters