Create_Titanic_Model.py

import pandas as pd
import numpy as np

from sklearn.model_selection import train_test_split
from sklearn.preprocessing import MinMaxScaler
from sklearn.linear_model import LogisticRegression

from sklearn.metrics import classification_report, confusion_matrix, accuracy_score

# load dataset
dataset = pd.read_csv('train.csv', encoding='latin-1')
dataset = dataset.rename(columns=lambda x: x.strip().lower())
dataset.head()

# cleaning missing values
dataset = dataset[['pclass', 'sex', 'age', 'sibsp', 'parch', 'fare', 'embarked', 'survived']]
dataset['sex'] = dataset['sex'].map({'male':0, 'female':1})
dataset['age'] = pd.to_numeric(dataset['age'], errors='coerce')
dataset['age'] = dataset['age'].fillna(np.mean(dataset['age']))

# dummy variables
embarked_dummies = pd.get_dummies(dataset['embarked'])
dataset = pd.concat([dataset, embarked_dummies], axis=1)
dataset = dataset.drop(['embarked'], axis=1)

X = dataset.drop(['survived'], axis=1)
y = dataset['survived']

# scaling features 
sc = MinMaxScaler(feature_range=(0,1))
X_scaled = sc.fit_transform(X)

# model fit
log_model = LogisticRegression(C=1)
log_model.fit(X_scaled, y)

# saving model as a pickle
import pickle
pickle.dump(log_model,open("titanic_survival_ml_model.sav", "wb"))
pickle.dump(sc, open("scaler.sav", "wb"))