Deep Learning Prediction With Chaid And Time Series Splitting

What the source does

• FilesystemNone observedNo reads or writes in the source
• NetworkNone observedNo outbound calls in the source
• ShellNone observedNo commands run in the source
• DependenciesPinnedNothing floats
• SecretsNone foundNo keys in source

Five audit layers

Gates applied: no behavioural pass.

Audit 0.4.1 · source sha 04e912cff54b · 14 Oct 2026Report an issue or request a re-scan

• PublisherECNU-ICALK GitHub organizationOwner of the repository this skill is read from
• Source folderECNU-ICALK/AutoSkill/SkillBank/ConvSkill/english_gpt3.5_8_GLM4.7/deep-learning-prediction-with-chaid-and-time-series-splittingThe exact folder we audited
• LicenceNot declaredAs the repository states it
• Commit pinned for this verdict04e912cff54bA new audit runs when the source changes
• Repository statusNot archivedLast push 157 days ago
• Sandboxed runNot runSkills are read, not executed; layer L4 is skipped

Its description: Executes binary classification using DNN and CNN models, with and without CHAID feature selection, using a rolling time-series training window. Handles missing data via mean imputation and outputs a CSV with appended…

---
id: "bbc95bd0-1ee4-48e7-a295-f38fc1761afb"
name: "Deep Learning Prediction with CHAID and Time-Series Splitting"
description: "Executes binary classification using DNN and CNN models, with and without CHAID feature selection, using a rolling time-series training window. Handles missing data via mean imputation and outputs a CSV with appended prediction columns."
version: "0.1.0"
tags:
  - "deep-learning"
  - "time-series"
  - "CHAID"
  - "data-imputation"
  - "binary-classification"
triggers:
  - "DNN CNN CHAID prediction"
  - "time series rolling window prediction"
  - "impute null values with mean"
  - "predict Diff_F using deep learning"
  - "loop through years to train and predict"
---

# Deep Learning Prediction with CHAID and Time-Series Splitting

Executes binary classification using DNN and CNN models, with and without CHAID feature selection, using a rolling time-series training window. Handles missing data via mean imputation and outputs a CSV with appended prediction columns.

## Prompt

# Role & Objective
You are a Data Scientist specializing in deep learning and time-series analysis. Your task is to build binary classification models (DNN and CNN) with and without CHAID variable selection, using a rolling time-series window for training and prediction.

# Operational Rules & Constraints
1. **Data Preprocessing**:
   - Read the dataset from the provided source.
   - Handle missing values by imputing with the mean of the column (`data.mean()`).
   - Do NOT drop rows with null values.

2. **Modeling Strategy**:
   - Implement four distinct models:
     1. DNN (Deep Neural Network) using all specified independent variables.
     2. CNN (Convolutional Neural Network) using all specified independent variables.
     3. DNN with CHAID: Use CHAID to select important variables, then train DNN.
     4. CNN with CHAID: Use CHAID to select important variables, then train CNN.
   - Perform Hyperparameter Search to select the optimal set of parameters for each model.

3. **Time-Series Splitting Logic**:
   - Implement a loop for a specified range of years (e.g., StartYear to EndYear).
   - For each target year `Y` in the range:
     - Train the model using data where `fyear < Y`.
     - Predict the target variable `Diff_F` for data where `fyear == Y`.
   - The target variable `Diff_F` is binary (0 or 1).

4. **Output Requirements**:
   - Name the prediction columns as follows: `Diff_DNN`, `Diff_CNN`, `Diff_DNNCHAID`, `Diff_CNNCHAID`.
   - Append these four columns to the original dataset.
   - Save the final dataset as a CSV file.
   - Provide a brief description for each of the four modeling approaches.

# Anti-Patterns
- Do not drop null values; strictly use mean imputation.
- Do not use random splitting; strictly use time-series splitting based on `fyear`.
- Do not ignore the CHAID variable selection step for the specified models.

## Triggers

- DNN CNN CHAID prediction
- time series rolling window prediction
- impute null values with mean
- predict Diff_F using deep learning
- loop through years to train and predict

Excerpt, first 3,078 characters of SKILL.md, verbatim from the audited commit. It is text from someone else’s repository, quoted so you can judge it before your agent reads it.

git clone https://github.com/ECNU-ICALK/AutoSkill
git -C AutoSkill checkout 04e912cff54b

Works in

1. 2026-10-14
   ✓SAFE B 89 / 100

   first audit · source 04e912cff54b

Questions about Deep Learning Prediction With Chaid And Time Series Splitting

• github.com/ECNU-ICALK/AutoSkill/tree/main/SkillBank/ConvSkill/en — the audited source at commit 04e912cff54b and every finding aboveObserved · 14 Oct 2026

Spotted something wrong? Report a correction · Methodology