Generate a Sklearn multiple-label classification model off of Automotive service appointments. The dataset was built from a proprietary dataset that was anonymized.

multiple-label-classification-appointments.py

"""
Labels automotive appointment description/comments
based on a trained multi-label classification model

Expected CSV structure:
ID|Tech|Service|Comments|mechanical|bodywork|diagnostic|suspension|engine|exhaust|electrical|brakes|tires

Structre notes:
ID, Tech, Service - Irrelevent, used during transcription
Comments - Used to generate the model
mechanical, bodywork, diagnostic, suspension, engine, exhaust, electrical, brakes, tires - Used to categorize (label)
"""

"""
Produced 2020
By https://amattu.com/links/github
Copy Alec M.
"""

"""
Original Author
https://www.geeksforgeeks.org/an-introduction-to-multilabel-classification/
"""

# Imports
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.feature_extraction.text import TfidfVectorizer
from skmultilearn.adapt import MLkNN

# Read model CSV
df = pd.read_csv('model.csv')
vetorizar = TfidfVectorizer(max_features = 3000, max_df = 0.85)
comments = ""

# Model text field
X = df["Comments"]
vetorizar.fit(X)

# Model data field(s)
y = np.asarray(df[df.columns[4:]])

# Split model datasets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.4)
X_train_tfidf = vetorizar.transform(X_train)
X_test_tfidf = vetorizar.transform(X_test)

# using Multi-label kNN classifier
mlknn_classifier = MLkNN()
mlknn_classifier.fit(X_train_tfidf, y_train)

# Continue reading until KeyboardInterupt
while comments == "":
    # Read appointment comments, test model
    comments = input("Enter a appointment description: ")
    new_sentence_tfidf = vetorizar.transform([comments])
    prediction = mlknn_classifier.predict(new_sentence_tfidf).toarray()[0]

    # Print results
    print("Mechanical: {}\nBodywork: {}\nDiagnostic: {}\nSuspension: {}\nEngine: {}\nExhaust: {}\nElectrical: {}\nBrakes: {}\nTires: {}".format(
        prediction[0],
        prediction[1],
        prediction[2],
        prediction[3],
        prediction[4],
        prediction[5],
        -1,  # untrained in dataset
        -1,  # untrained in dataset
        -1   # untrained in dataset
    ))

    # Reset comments
    comments = ""

Dataset not public ATM, currently scraping confidential data out of it.