/PredictiveMaintenance

A data science personal project aiming to predict machine failure on imbalance data by using classification algorithms

Primary LanguageJupyter Notebook

Predictive Maintenance

Background

Maintaining the availability of machines is something that needs to be considered for the continuity of activities at a company. Maintenance is carried out to avoid machine failure during production activities. Machine failure at the company will bring losses to the company.

In this Idustry 4.0, many companies implement big data analytics to predict in predicting machine failure. Predictive maintenance works better than corrective or preventive maintenance. It can continuously predict future failures and any Remaining Useful Life (RUL) of the equipment.

Goals

As data scientist, We can create predictive model for diagnostic of machine failure and anlyze which factors that bring machine failure based on the dataset.

Dataset

The dataset can be downloaded from archive.ics.uci.edu . The file used is ai4i2020.csv containing 10000 rows and 14 columns.

Steps

  1. Data Preparation
  2. Exploratory Data Analysis
  3. Data Pre-Processing
    • Handling Outliers
    • Scaling Data
    • One-Hot Encoding
    • Resampling
  4. Modeling
    • Logistic Regression
    • K-Nearest Neighbor Classifier
    • Random Forest Classifier
  5. Model Evaluation & Model Comparison