ium_464953/MLProject/model_creator.py

92 lines
3.5 KiB
Python
Raw Normal View History

2024-05-13 23:32:25 +02:00
import mlflow
import pandas as pd
import os
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score
from sklearn.preprocessing import StandardScaler
from sklearn. preprocessing import LabelEncoder
import pickle
import sys
def datasets_preparation():
df_1 = pd.read_csv("spotify_songs.csv")
df_2 = pd.read_csv("Spotify_Dataset.csv", sep=";")
df_1 = df_1.dropna()
df_2 = df_2.dropna()
df_2 = df_2.rename(columns={'Title': 'track_name'})
columns_to_remove_df_1 = ['track_id', 'track_album_id', 'track_album_name', 'track_album_release_date',
'playlist_id', 'playlist_subgenre']
columns_to_remove_df_2 = ['Date','# of Artist', 'Artist (Ind.)', '# of Nationality',
'Nationality', 'Continent', 'Points (Total)',
'Points (Ind for each Artist/Nat)', 'id', 'Song URL']
df_1 = df_1.drop(columns=columns_to_remove_df_1)
df_2 = df_2.drop(columns=columns_to_remove_df_2)
df_1 = df_1.drop_duplicates(subset=['track_name'])
df_2 = df_2.drop_duplicates(subset=['track_name'])
le = LabelEncoder()
unique_names_df2 = df_2['track_name'].unique()
diff_df = df_1[~df_1['track_name'].isin(unique_names_df2)]
diff_df = diff_df.iloc[:10000]
diff_df['track_artist'] = le.fit_transform(diff_df.track_artist)
diff_df['playlist_name'] = le.fit_transform(diff_df.playlist_name)
diff_df['playlist_genre'] = le.fit_transform(diff_df.playlist_genre)
if "docker_test_dataset.csv" not in os.listdir():
diff_df.to_csv("docker_test_dataset.csv", index=False)
result_df = pd.merge(df_1, df_2, on='track_name', how='inner')
result_df = result_df.drop_duplicates(subset=['track_name'])
columns_to_remove_result_df = ['Rank', 'Artists', 'Danceability', 'Energy', 'Loudness',
'Speechiness', 'Acousticness', 'Instrumentalness', 'Valence']
result_df = result_df.drop(columns=columns_to_remove_result_df)
result_df['track_artist'] = le.fit_transform(result_df.track_artist)
result_df['playlist_name'] = le.fit_transform(result_df.playlist_name)
result_df['playlist_genre'] = le.fit_transform(result_df.playlist_genre)
return result_df
def train_model_and_log(max_iter):
result_df = datasets_preparation()
Y = result_df[['playlist_genre']]
X = result_df.drop(columns='playlist_genre')
X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.10, random_state=42)
Y_train = np.ravel(Y_train)
Y_test = np.ravel(Y_test)
scaler = StandardScaler()
numeric_columns = X_train.select_dtypes(include=['int', 'float']).columns
X_train_scaled = scaler.fit_transform(X_train[numeric_columns])
X_test_scaled = scaler.transform(X_test[numeric_columns])
model = LogisticRegression(max_iter=max_iter)
model.fit(X_train_scaled, Y_train)
Y_pred = model.predict(X_test_scaled)
accuracy = accuracy_score(Y_test, Y_pred)
print("Accuracy:", accuracy)
with mlflow.start_run():
mlflow.log_param("model_type", "LogisticRegression")
mlflow.log_param("test_size", 0.1)
mlflow.log_param("max_iter", max_iter)
mlflow.log_metric("accuracy", accuracy)
mlflow.sklearn.log_model(model, "model")
if __name__ == "__main__":
max_iter = int(sys.argv[1]) if len(sys.argv) > 1 else 1000
mlflow.set_tracking_uri("http://localhost:5000")
train_model_and_log(max_iter)