MACHINE MAINTENANCE ANALYSIS

Group 3 analyzed a synthetic dataset for determining machine failures. Six features were provided, while a multivariate target included failure types. The end result was that the feature "Tool Wear" had the biggest impact on machine failure, though various features contribute to different failure types. A company in this situation should develop a preventative maintenance plan to mitigate these issues in the future.

---

The team approached this analysis in the following manner:

1. Data overview
2. Analyses and conclusions
3. Next steps

---

The following sections break down the overall approach:

1. Data Overview
   • 10,000 rows x 10 columns
   • IDs
     • UID
       • Unique identifier ranging from 1 to 10000
     • ProductID
       • Consisting of a letter L, M, or H for low (50% of all products), medium (30%), or high (20%) as product quality variants and a variant-specific serial number
   • Features (X)
     • Type
       • Consisting of a letter L, M, or H for low, medium, or high
     • Air temperature [K]
       • Generated using a random walk process later normalized to a standard deviation of 2 K around 300 K
     • Process temperature [K]
       • Generated using a random walk process normalized to a standard deviation of 1 K, added to the air temperature plus 10 K
     • Rotational speed [rpm]
       • Calculated from powepower of 2860 W, overlaid with a normally distributed noise
     • Torque [Nm]
       • Torque values are normally distributed around 40 Nm with an σ = 10 Nm and no negative values
     • Tool wear [min]
       • The quality variants H/M/L add 5/3/2 minutes of tool wear to the used tool in the process. and a 'machine failure' label that indicates, whether the machine has failed in this particular data point for any of the following failure modes are true
   • Targets (Y)
     • Target
       • 0 (No Failure)
       • 1 (Failure)
     • Failure Type
       • Heat Dissipation Failure
       • Overstrain Failure
       • No Failure
       • Power Failure
       • Random Failures
       • Tool Wear Failure
2. Analyses and Conclusions
   • Data Cleaning and Manipulation
     • Dropped ID columns (e.g., UID, ProductID)
     • Converted feature "Failure Type" into dummy variables
     • Split the data into training and test sets
     • Scaled the data in the feature columns
     • Used one-hot encoding for feature "Type" (e.g., L, M, H)
   • Pipeline
     • xxx
   • Models
     • xxx
   • xxx
     • xxx

• Accuracy Scores

  Model	As-Is Data	Oversampled Data	Undersampled Data
  Logistic Regression	xxx	xxx	xxx
  K-Nearest Neighbors	xxx	xxx	xxx
  Decision Tree	xxx	xxx	xxx
  Random Forest	xxx	xxx	xxx
  Extremely Random Trees	xxx	xxx	xxx
  Gradient Boosting	xxx	xxx	xxx
  AdaBoost	xxx	xxx	xxx
  Support Vector Machine	xxx	xxx	xxx
  Naive Bayes	xxx	xxx	xxx
  xxx	xxx	xxx	xxx

3. Next Steps
   • Develop a preventative maintanence plan
   • xxx
   • xxx

Resources

• Kaggle: https://www.kaggle.com/datasets/shivamb/machine-predictive-maintenance-classification

---