import numpy as np import matplotlib.pyplot as plt %matplotlib inline import seaborn as sns import pandas as pd import scipy.stats as stats import plotly.express as px #import graphviz import pylab as pl from sklearn import preprocessing, metrics from sklearn.model_selection import train_test_split from sklearn.dummy import DummyClassifier, DummyRegressor from sklearn.linear_model import LinearRegression, Ridge, BayesianRidge from sklearn.neighbors import KNeighborsClassifier #from sklearn.tree import DecisionTreeClassifier, export_graphviz from sklearn.metrics import accuracy_score,mean_squared_error, r2_score, classification_report, confusion_matrix from sklearn.cluster import KMeans from sklearn.decomposition import PCA from sklearn.mixture import GaussianMixture from mpl_toolkits.mplot3d import Axes3D from mpl_toolkits import mplot3d

movie_df = pd.read_csv('movie.csv') movie_df.head(1)

extern_movies_df = pd.read_csv('movies_extern.csv', sep = ';') extern_movies_df.head(1)

movie_df.isnull().sum()

movie_df['language'].unique()

movie_df['country'].unique()

movie_df['facenumber_in_poster'].sort_values().unique()

movie_df = movie_df[ [ 'imdb_score', 'gross', 'genres', 'movie_title', 'cast_total_facebook_likes', 'facenumber_in_poster', 'country', 'title_year', 'budget', 'movie_facebook_likes', 'director_facebook_likes', 'duration', 'content_rating', 'language', 'director_name' ] ] movie_df.head(1)

movie_df = movie_df.replace(['\xa0', '\n', '\t','"'], '', regex = True) movie_df.movie_title = movie_df.movie_title.astype("string")

extern_movies_df.rename({'title': 'movie_title'}, axis=1, inplace = True) # Kolomnaam veranderen zodat hij overeenkomt met onze orginele dataset extern_movies_df.movie_title = extern_movies_df.movie_title.astype("string")

movie_df = pd.merge(movie_df, extern_movies_df[['movie_title','revenue', 'budget', 'vote_average', 'vote_count', 'popularity']], how='inner', on='movie_title', suffixes=('_original', '_extern')) movie_df.drop_duplicates(subset = 'movie_title', inplace = True)

movie_df.dtypes

dummies_genres = movie_df.genres.str.get_dummies() movie_df = movie_df.join(dummies_genres) movie_df.head(1)

movie_df.facenumber_in_poster.isnull().sum()

movie_df.country = movie_df.country.astype('category')

movie_df.title_year = movie_df.title_year.astype(int)

movie_df.budget_original.isnull().sum()

movie_df.director_facebook_likes = movie_df.director_facebook_likes.astype(int) movie_df.movie_facebook_likes = movie_df[movie_df['title_year'] > 2006].movie_facebook_likes.replace(0,np.nan) movie_df.cast_total_facebook_likes = movie_df[movie_df['title_year'] > 2006].cast_total_facebook_likes.replace(0,np.nan) movie_df.director_facebook_likes = movie_df[movie_df['title_year'] > 2006].director_facebook_likes.replace(0,np.nan)

movie_df.duration.isna().sum()

movie_df.content_rating = movie_df.content_rating.astype('category')

movie_df.language = movie_df.language.astype('category')

movie_df.vote_count.plot.box()

print("Voor het verwijderen van de outliers") print(movie_df.vote_count.describe()) q1 = movie_df.vote_count.quantile(0.25) q3 = movie_df.vote_count.quantile(0.75) iqr = q3 - q1 # Zwakke outliers zwakke_outliers = movie_df[((movie_df['vote_count'] < q1 - 1.5 * iqr) & (movie_df['vote_count'] > q1 - 3 * iqr)) | ((movie_df['vote_count'] < q1 + 3 * iqr) & (movie_df['vote_count'] > q1 + 1.5 * iqr)) ] # Sterke outliers movie_df[(movie_df['vote_count'] < q1 - 3 * iqr) | (movie_df['vote_count'] > q3 + 3 * iqr)] = np.nan print("\nNa het verwijderen van de outliers") print(movie_df.vote_count.describe())

movie_df.vote_count.plot.box()

movie_df = movie_df[pd.to_numeric(movie_df['popularity'], errors='coerce').notnull()] movie_df.popularity = movie_df.popularity.dropna().astype(float) movie_df.popularity.describe()

movie_df.dtypes

res_1_df = movie_df[ [ 'movie_title', 'imdb_score', 'gross', 'budget_original', 'budget_extern', 'movie_facebook_likes', 'cast_total_facebook_likes', 'director_facebook_likes', 'title_year', 'vote_average', 'vote_count', 'popularity', 'revenue' ] ].copy() res_1_df.describe()

res_1_df.dropna(subset=['gross', 'revenue', 'budget_original', 'budget_extern', 'imdb_score', 'vote_average', 'vote_count', 'popularity'], inplace=True) res_1_df.dropna(subset=['movie_facebook_likes', 'director_facebook_likes', 'cast_total_facebook_likes'], inplace=True, how='all')

res_1_df['movie_facebook_likes'].plot.hist(bins=20).set_title('frequency van de movie facebook likes')

res_1_df[['movie_facebook_likes', 'cast_total_facebook_likes', 'director_facebook_likes']] = res_1_df[['movie_facebook_likes', 'cast_total_facebook_likes', 'director_facebook_likes']].fillna(0)

res_1_df['total_facebook_likes'] = res_1_df['movie_facebook_likes'] + res_1_df['cast_total_facebook_likes'] + res_1_df['director_facebook_likes'] res_1_df.total_facebook_likes = res_1_df[res_1_df['title_year'] > 2006].total_facebook_likes.replace(0,np.nan) res_1_df.dropna(subset=['total_facebook_likes'],inplace=True) res_1_df.total_facebook_likes = res_1_df.total_facebook_likes.astype(int)

fig, axes = plt.subplots(nrows=1,ncols=2,figsize=(12,4)) res_1_df[['imdb_score']].plot.hist(bins=15, ax=axes[0], subplots=True) res_1_df[['imdb_score']].plot.box(ax=axes[1], subplots=True)

sns.scatterplot(x='imdb_score', y='total_facebook_likes', size='gross', hue='gross', data=res_1_df[['imdb_score', 'total_facebook_likes', 'gross']])

corrMatrix2 = res_1_df.corr() fig, axes = plt.subplots(figsize=(10,10)) sns.heatmap(corrMatrix2, annot=True,ax=axes)

fig, axes = plt.subplots(2,2,figsize=(20,10)) sns.scatterplot(x='imdb_score', y='gross', hue='gross', data=res_1_df[['imdb_score', 'gross']],ax=axes[0][0]) axes[0][0].set_title('Scatterplot van imbd score en gross') sns.scatterplot(x='movie_facebook_likes', y='gross', hue='gross', data=res_1_df[['movie_facebook_likes', 'gross']],ax=axes[0][1]) axes[0][1].set_title('Scatterplot van de films FB likes en gross') sns.scatterplot(x='total_facebook_likes', y='gross', hue='gross', data=res_1_df[['total_facebook_likes', 'gross']], ax=axes[1][0]) axes[1][0].set_title('Scatterplot van alle FB likes en gross')

x_research_1_A = res_1_df[['imdb_score', 'total_facebook_likes']] y_research_1_A = res_1_df[['gross']] x_research_1_A_train, x_research_1_A_test, y_research_1_A_train, y_research_1_A_test = train_test_split(x_research_1_A, y_research_1_A, random_state=0)

x_research_1_B = res_1_df[['budget_extern', 'vote_count']] y_research_1_B = res_1_df[['revenue']] x_research_1_B_train, x_research_1_B_test, y_research_1_B_train, y_research_1_B_test = train_test_split(x_research_1_B, y_research_1_B, random_state=0)

print(f'Totaal A: {len(movie_df)}, 75%: {len(x_research_1_A_train)}, 25%: {len(y_research_1_A_test)}') print(f'Totaal B: {len(movie_df)}, 75%: {len(x_research_1_B_train)}, 25%: {len(y_research_1_B_test)}')

y_baseline_1_A_means = pd.DataFrame({'means' : np.repeat(y_research_1_A_train.gross.mean(), y_research_1_A_train.shape[0])}) baseline_1_A_RMSE_score = np.sqrt(mean_squared_error(y_research_1_A_train, y_baseline_1_A_means)) baseline_1_A_RMSE_score

y_baseline_1_B_means = pd.DataFrame({'means' : np.repeat(y_research_1_B_train.revenue.mean(), y_research_1_B_train.shape[0])}) baseline_1_B_RMSE_score = np.sqrt(mean_squared_error(y_research_1_B_train, y_baseline_1_B_means)) baseline_1_B_RMSE_score

print(x_research_1_A_train.shape) print(y_research_1_A_test.shape)

linreg_research_1_A = LinearRegression() linreg_research_1_A.fit(x_research_1_A_train, y_research_1_A_train)

y_research_1_A_pred = linreg_research_1_A.predict(x_research_1_A_test)

linreg_1_A_RMSE_score = np.sqrt(mean_squared_error(y_research_1_A_test, y_research_1_A_pred)) linreg_1_A_RMSE_score

linreg_1_A_r2_score = r2_score(y_research_1_A_test, y_research_1_A_pred) linreg_1_A_r2_score

linreg_research_1_B = LinearRegression() linreg_research_1_B.fit(x_research_1_B_train, y_research_1_B_train)

y_research_1_B_pred = linreg_research_1_B.predict(x_research_1_B_test)

linreg_1_B_RMSE_score = np.sqrt(mean_squared_error(y_research_1_B_test, y_research_1_B_pred)) linreg_1_B_RMSE_score

linreg_1_B_r2_score = r2_score(y_research_1_B_test, y_research_1_B_pred) linreg_1_B_r2_score

ridge_research_1_A = Ridge() ridge_research_1_A.fit(x_research_1_A_train, y_research_1_A_train)

y_ridge_research_1_A_pred = ridge_research_1_A.predict(x_research_1_A_test)

ridge_1_A_RMSE_score = np.sqrt(mean_squared_error(y_research_1_A_test, y_ridge_research_1_A_pred)) ridge_1_A_RMSE_score

ridge_1_A_r2_score = r2_score(y_research_1_A_test, y_ridge_research_1_A_pred) ridge_1_A_r2_score

ridge_research_1_B = Ridge() ridge_research_1_B.fit(x_research_1_B_train, y_research_1_B_train)

y_ridge_research_1_B_pred = ridge_research_1_B.predict(x_research_1_B_test)

ridge_1_B_RMSE_score = np.sqrt(mean_squared_error(y_research_1_B_test, y_ridge_research_1_B_pred)) ridge_1_B_RMSE_score

ridge_1_B_r2_score = r2_score(y_research_1_B_test, y_ridge_research_1_B_pred) ridge_1_B_r2_score

fig = px.scatter_3d(res_1_df, x='imdb_score', y='total_facebook_likes', z='gross') fig.show()

fig = px.scatter_3d(res_1_df, x='budget_extern', y='vote_count', z='revenue') fig.show()

print(f"σ van gross : {res_1_df.gross.std().round()}") print(f"RSME van Baseline A: {baseline_1_A_RMSE_score.round()}") print(f"RSME van Lineare Regressie A: {linreg_1_A_RMSE_score.round()}") print(f"R2 van Lineare Regressie A: {linreg_1_A_r2_score.round(2)}") print('\n') print(f"RSME van Ridge Regressie A: {ridge_1_A_RMSE_score.round()}") print(f"R2 van Ridge Regressie A: {ridge_1_A_r2_score.round(2)}") print('\n') print('\n') print(f"σ van revenue : {res_1_df.revenue.std().round()}") print(f"RSME van Baseline B: {baseline_1_B_RMSE_score.round()}") print(f"RSME van Lineare Regressie B: {linreg_1_B_RMSE_score.round()}") print(f"R2 van Lineare Regressie B: {linreg_1_B_r2_score.round(2)}") print('\n') print(f"RSME van Ridge Regressie B: {ridge_1_B_RMSE_score.round()}") print(f"R2 van Ridge Regressie B: {ridge_1_B_r2_score.round(2)}")

res_2_df = movie_df[ [ 'duration','title_year', 'budget_original', 'budget_extern','content_rating', 'language', 'country', 'genres', 'Action', 'Adventure','Animation', 'Biography','Comedy','Crime','Documentary', 'Drama','Family','Fantasy','Film-Noir','History', 'Horror','Music','Musical', 'Mystery','News','Romance', 'Sci-Fi','Sport','Thriller', 'War','Western', ] ].copy() res_2_df.describe()

temp = res_2_df.isna().sum() temp = temp[temp>0] temp

# drop alles NaNs bij res_2_df.dropna(subset=['duration', 'budget_original','content_rating','language'], inplace=True) print(f"Nan's bij duration. Na de drop commando:",res_2_df.duration.isna().sum()) print(f"Nan's bij budget_original. Na de drop commando:",res_2_df.budget_original.isna().sum()) print(f"Nan's bij content_rating. Na de drop commando:",res_2_df.content_rating.isna().sum()) print(f"Nan's bij language. Na de drop commando:",res_2_df.language.isna().sum())

# get dummies for language, content rating en country for column in ['content_rating', 'language', 'country']: temp = res_2_df[column].str.get_dummies() res_2_df = res_2_df.join(temp) res_2_df.head(1)

kolommen = ['duration','title_year','budget_original','budget_extern'] for enkele_kolom in kolommen: kolom = res_2_df[enkele_kolom].sort_values() Q1 = kolom.quantile(0.25) Q3 = kolom.quantile(0.75) IQR = Q3 - Q1 temp_zwakke_uitschieters_lower = kolom[(kolom <= Q1 - 1.5*IQR) & (kolom > Q1 - 3*IQR)].count() temp_zwakke_uitschieters_upper = kolom[(kolom >= Q3 + 1.5*IQR) & (kolom < Q3 + 3*IQR)].count() temp_sterke_uitschieters_lower = kolom[kolom < Q1 - 3*IQR].count() temp_sterke_uitschieters_upper = kolom[kolom > Q3 + 3*IQR].count() print(f"{enkele_kolom} zwakke outliers lager: {temp_zwakke_uitschieters_lower}") print(f"{enkele_kolom} sterke outliers lager: {temp_sterke_uitschieters_lower}") print(f"{enkele_kolom} zwakke outliers hoger: {temp_zwakke_uitschieters_upper}") print(f"{enkele_kolom} sterke outliers hoger: {temp_sterke_uitschieters_upper}")

fig,axes = plt.subplots(1,4,figsize=(20,10)) res_2_df.boxplot(column=kolommen[0],ax=axes[0]) res_2_df.boxplot(column=kolommen[1],ax=axes[1]) res_2_df.boxplot(column=kolommen[2],ax=axes[2]) res_2_df.boxplot(column=kolommen[3],ax=axes[3])

corrMatrix_1 = res_2_df[ [ 'title_year', 'duration', 'budget_original', 'budget_extern' ] ].corr() corrMatrix_2 = res_2_df[ [ 'duration', 'Action', 'Adventure','Animation', 'Biography','Comedy','Crime','Documentary', 'Drama','Family','Fantasy','Film-Noir','History', 'Horror','Music','Musical', 'Mystery','News','Romance', 'Sci-Fi','Sport','Thriller', 'War','Western' ] ].corr() # experimentele kolommen voor later corrMatrix_3 = res_2_df[ [ 'duration','Approved','G' ,'GP','M','NC-17','Not Rated','PG','PG-13','Passed','R','TV-14','TV-G','Unrated','X' ] ].corr() corrMatrix_4 = res_2_df[ [ 'duration','English', 'Mandarin', 'Aboriginal', 'Spanish','French', 'Filipino', 'Russian', 'Maya', 'Kazakh', 'Cantonese', 'Japanese', 'Aramaic', 'Italian', 'Dutch', 'Dari', 'German', 'Mongolian', 'Bosnian', 'Korean', 'Hungarian', 'Hindi', 'Thai', 'Portuguese', 'Norwegian', 'Czech', 'Danish', 'Zulu', 'Hebrew', 'Arabic', 'Vietnamese', 'Indonesian', 'Romanian', 'Persian' ] ].corr() corrMatrix_5 = res_2_df[ [ 'duration', 'UK', 'USA', 'New Zealand', 'Australia', 'Germany', 'New Line', 'France', 'China', 'Canada', 'Mexico', 'Spain', 'Hong Kong', 'Czech Republic', 'Soviet Union', 'South Korea', 'Peru', 'Italy', 'Japan', 'Aruba', 'Denmark', 'Libya', 'Ireland', 'Romania', 'West Germany', 'South Africa', 'Chile', 'Russia', 'Netherlands', 'Hungary', 'Panama', 'Belgium', 'Greece', 'Taiwan', 'Bulgaria', 'Iran', 'Georgia', 'India', 'Thailand', 'Norway', 'Philippines', 'Brazil', 'Finland', 'Bahamas', 'Iceland', 'Argentina', 'Colombia', 'Poland', 'Israel', 'Egypt', 'Kyrgyzstan', 'Afghanistan', 'Switzerland' ] ].corr() fig,axes = plt.subplots(5,1,figsize=(20,50)) sns.heatmap(corrMatrix_1, annot=True,ax=axes[0]) sns.heatmap(corrMatrix_2, annot=True,ax=axes[1]) sns.heatmap(corrMatrix_3, annot=True,ax=axes[2]) sns.heatmap(corrMatrix_4, annot=True,ax=axes[3]) sns.heatmap(corrMatrix_5, annot=True,ax=axes[4])

X_baseline_2 = res_2_df[ [ 'title_year', 'budget_original', 'budget_extern', 'Action', 'Adventure','Animation', 'Biography','Comedy','Crime','Documentary', 'Drama','Family','Fantasy','Film-Noir','History', 'Horror','Music','Musical', 'Mystery','News','Romance', 'Sci-Fi','Sport','Thriller', 'War','Western' ] ] Y_baseline_2 = res_2_df[['duration']]

baseline_2_DummyRegre = DummyRegressor(strategy='mean')

baseline_2_DummyRegre.fit(X_baseline_2,Y_baseline_2)

Dummy_2_pred = baseline_2_DummyRegre.predict(X_baseline_2)

Dummyreg_2_RMSE_score = np.sqrt(mean_squared_error(Y_baseline_2, Dummy_2_pred)) Dummyreg_2_RMSE_score

baseline_2_DummyRegre_score = baseline_2_DummyRegre.score(Y_baseline_2,Y_baseline_2) print(baseline_2_DummyRegre_score)

x_research_2_train, x_research_2_test, y_research_2_train, y_research_2_test = train_test_split(X_baseline_2, Y_baseline_2, random_state=0)

print(x_research_2_train.shape) print(y_research_2_test.shape)

linreg_research_2 = LinearRegression() linreg_research_2.fit(x_research_2_train, y_research_2_train)

y_research_2_pred = linreg_research_2.predict(x_research_2_test)

linreg_2_RMSE_score = np.sqrt(mean_squared_error(y_research_2_test, y_research_2_pred)) linreg_2_RMSE_score

research_2_liniear_r2_score = r2_score(y_research_2_test,y_research_2_pred) research_2_liniear_r2_score

ridge_research_2 = Ridge() ridge_research_2.fit(x_research_2_train, y_research_2_train)

y_ridge_research_2_pred = ridge_research_2.predict(x_research_2_test)

ridge_2_RMSE_score = np.sqrt(mean_squared_error(y_research_2_test, y_ridge_research_2_pred)) ridge_2_RMSE_score

research_2_ridge_r2_score = r2_score(y_research_2_test,y_ridge_research_2_pred) research_2_ridge_r2_score

smaller_data_pool_experiment_x = res_2_df[ [ 'Drama', 'History','War','Biography' ] ]

x_experiment_2_train, x_experiment_2_test, y_experiment_2_train, y_experiment_2_test = train_test_split(smaller_data_pool_experiment_x, Y_baseline_2, random_state=0)

linreg_experiment = LinearRegression() linreg_experiment.fit(x_experiment_2_train, y_experiment_2_train) ridge_experiment = Ridge() ridge_experiment.fit(x_experiment_2_train, y_experiment_2_train)

y_linear_experiment_2_pred = linreg_experiment.predict(x_experiment_2_test) y_ridge_experiment_2_pred = ridge_experiment.predict(x_experiment_2_test)

linreg_2_experiment_RMSE_score = np.sqrt(mean_squared_error(y_experiment_2_test, y_linear_experiment_2_pred)) linreg_2_experiment_RMSE_score

ridge_2_experiment_RMSE_score = np.sqrt(mean_squared_error(y_experiment_2_test, y_ridge_experiment_2_pred)) ridge_2_experiment_RMSE_score

experiment_2_liniear_r2_score = r2_score(y_experiment_2_test,y_linear_experiment_2_pred) experiment_2_liniear_r2_score

experiment_2_ridge_r2_score = r2_score(y_experiment_2_test,y_ridge_experiment_2_pred) experiment_2_ridge_r2_score

print('Het resultaat van de Baseline,Linear en Ridge model met de niet optionele kolommen') print(f"Accuracy van Baseline : {baseline_2_DummyRegre_score}") print(f"Accuracy van Linear : {research_2_liniear_r2_score}") print(f"Accuracy van Ridge : {research_2_ridge_r2_score}") print('\n') print("Het resultaat van de data experiment") print(f"Accuracy van Linear : {experiment_2_liniear_r2_score}") print(f"Accuracy van Ridge : {experiment_2_ridge_r2_score}")

res_3_df = movie_df[ [ 'budget_extern', 'revenue', 'director_name' ] ].copy() res_3_df.head(5)

res_3_df = res_3_df.replace(0, np.nan) res_3_df = res_3_df.dropna(how='any', axis=0) res_3_df

res_3_df["winstmarge"] = ((res_3_df.revenue - res_3_df.budget_extern) / res_3_df.revenue * 100).round(2) res_3_df

kolom = res_3_df['winstmarge'].sort_values() Q1 = kolom.quantile(0.25) Q3 = kolom.quantile(0.75) IQR = Q3 - Q1 winstmarge_zwakke_uitschieters_lower = kolom[(kolom <= Q1 - 1.5*IQR) & (kolom > Q1 - 3*IQR)].count() winstmarge_zwakke_uitschieters_upper = kolom[(kolom >= Q3 + 1.5*IQR) & (kolom < Q3 + 3*IQR)].count() winstmarge_sterke_uitschieters_lower = kolom[kolom < Q1 - 3*IQR].count() winstmarge_sterke_uitschieters_upper = kolom[kolom > Q3 + 3*IQR].count() print(f"winstmarge zwakke outliers lager: {winstmarge_zwakke_uitschieters_lower}") print(f"winstmarge sterke outliers lager: {winstmarge_sterke_uitschieters_lower}") print(f"winstmarge zwakke outliers hoger: {winstmarge_zwakke_uitschieters_upper}") print(f"winstmarge sterke outliers hoger: {winstmarge_sterke_uitschieters_upper}") res_3_df.boxplot(column='winstmarge')

res_3_df = res_3_df[res_3_df['winstmarge'] > -100] res_3_df

amount_movies = res_3_df.groupby('director_name', as_index=False).size()

res_3_df = pd.DataFrame(res_3_df.groupby('director_name', as_index=False).agg({'winstmarge': "mean"})) res_3_df.head(1)

res_3_df = res_3_df.merge(amount_movies, on=['director_name'])

res_3_df.rename(columns={'size':'amount_movies'}, inplace=True)

res_3_df.director_name = res_3_df.director_name.astype('category') le = preprocessing.LabelEncoder() le.fit(res_3_df['director_name'].astype(str)) res_3_df['director_name_int'] = le.transform(res_3_df['director_name'])

res_3_df

kmeans_res_3 = KMeans(n_clusters=5, random_state=0)

kmeans_3_scores = [] for k in range(1, 16): kmeans_res_3.set_params(n_clusters=k, random_state=0) kmeans_res_3.fit(res_3_df[['winstmarge', 'amount_movies']]) kmeans_3_scores.append(kmeans_res_3.score(res_3_df[['winstmarge', 'amount_movies']])) temp_x = np.arange(1,16,step=1) plt.title('Elbow methode voor bepalen van K') plt.plot(temp_x,kmeans_3_scores)

kmeans_res_3.set_params(n_clusters=5, random_state=0) kmeans_res_3.fit(res_3_df[['winstmarge', 'amount_movies']]) kmeans_res_3.labels_ kmeans_res_3.score(res_3_df[['winstmarge', 'amount_movies']])

model_3_score = metrics.silhouette_score(res_3_df[['winstmarge', 'amount_movies']], kmeans_res_3.labels_, metric='euclidean') print(model_3_score)

res_3_df

res_3_df['clusters'] = kmeans_res_3.labels_

sns.scatterplot(x='winstmarge', y='amount_movies', size='clusters', hue='clusters', data=res_3_df[['winstmarge', 'amount_movies', 'clusters']])

print(f"accuracy score van het model: {kmeans_3_scores[4]}") print(f"silouette score van het model: {model_3_score}")