Machine learning by SPSS

Here’s a step-by-step example of applying Machine Learning (ML) using SPSS to classify participants into groups based on psychological data:

---

Example Project: Classifying Participants into Stress Levels

Objective

To use SPSS to predict whether participants fall into "High Stress" or "Low Stress" groups based on survey responses.

Dataset

Imagine you have a dataset with the following columns:

1. Age (numerical)
2. Hours of Sleep per Night (numerical)
3. Number of Stressful Events in the Last Month (numerical)
4. Self-Reported Stress Level (categorical: "High" or "Low")

---

Steps in SPSS

1. Load the Data

• Open SPSS.
• Import the dataset (e.g., in .sav or .csv format).
• Ensure columns are labeled correctly and data is clean.

2. Data Preparation

• Recode variables if needed: For example, ensure "High" and "Low" in the stress level column are coded as 1 and 0 for binary classification.
• Check for missing values and handle them (e.g., replace with mean/median or exclude cases).
• Normalize numerical variables if needed (optional).

3. Exploratory Data Analysis (EDA)

• Use Descriptive Statistics to understand the data distribution.
• Create Boxplots or Scatterplots to explore relationships (e.g., stress vs. sleep hours).

4. Train a Decision Tree Model

1. Navigate to Analyze > Classify > Decision Trees.
2. Set the dependent variable to "Stress Level" (High/Low).
3. Set the independent variables to Age, Sleep Hours, and Stressful Events.
4. Choose the CHAID or CART algorithm for the decision tree.
5. Configure stopping criteria (e.g., minimum number of cases per node).
6. Click OK to run the model.

5. Interpret the Results

• SPSS generates a decision tree diagram showing how variables split to classify participants.
• For example:
  • If "Stressful Events > 5," classify as "High Stress."
  • Otherwise, check "Hours of Sleep < 6" to classify as "High Stress."

6. Evaluate the Model

• Look at the classification table to assess accuracy, sensitivity, and specificity.
• Analyze the confusion matrix to see how well the model predicts "High" vs. "Low" stress.

7. Use the Model for Prediction

• Apply the model to a new dataset using the Scoring Wizard in SPSS.
• Generate predicted stress levels for new participants based on input features.

---

Outcome

• Students learn to build a simple decision tree model in SPSS.
• They interpret results and understand practical applications of machine learning in psychology.
• Real-world application: Predicting stress levels for psychological interventions.