ARTIFICIAL INTELLIGENCE PRACTICAL FILE
AI-Based Growth / Disease Risk Prediction
Student: Ratnadip Ghosh | Class XII Humanities | KV No.2 Kanchrapara
Project Introduction
This practical file documents the full AI lifecycle including preprocessing, model training, evaluation, anomaly
detection, and automated reporting.
Figure: Visualization generated to support model interpretation.
Setup & Imports
These libraries enable machine learning, evaluation, and visualization.
import pandas as pd
import numpy as np
import [Link] as plt
from sklearn.model_selection import train_test_split
from [Link] import RandomForestClassifier
from [Link] import LabelEncoder
from [Link] import accuracy_score, confusion_matrix, classification_report
Figure: Visualization generated to support model interpretation.
Load Dataset
The dataset is loaded and categorical columns are encoded for model compatibility.
df = pd.read_csv("merged_ai_dataset.csv")
le_id = LabelEncoder()
le_y = LabelEncoder()
df["ID_risk"] = le_id.fit_transform(df["ID_risk"])
df["Growth_Status_enc"] = le_y.fit_transform(df["Growth_Status"])
Figure: Visualization generated to support model interpretation.
Feature Selection
Relevant predictors are selected to improve training efficiency.
X = df[["Age_months","Height_z","Weight_z","BMI_z","ASD_score","ID_risk"]]
y = df["Growth_Status_enc"]
Figure: Visualization generated to support model interpretation.
Train-Test Split
Splitting ensures unbiased evaluation.
X_train, X_test, y_train, y_test = train_test_split(
X, y, test_size=0.2, random_state=42, stratify=y
)
Figure: Visualization generated to support model interpretation.
Model Training
Random Forest builds multiple trees and aggregates predictions.
model = RandomForestClassifier(n_estimators=300, random_state=42)
[Link](X_train, y_train)
Figure: Visualization generated to support model interpretation.
Prediction
The trained model predicts unseen data.
y_pred = [Link](X_test)
Figure: Visualization generated to support model interpretation.
Evaluation
Accuracy and confusion matrix validate performance.
acc = accuracy_score(y_test, y_pred)
cm = confusion_matrix(y_test, y_pred)
print("Accuracy:", acc)
Figure: Visualization generated to support model interpretation.
Anomaly Detection
Z-score rule identifies abnormal growth patterns.
df["Anomaly_Flag"] = ((df["Height_z"].abs() > 3) |
(df["Weight_z"].abs() > 3)).astype(int)
Figure: Visualization generated to support model interpretation.
Feature Importance
Explains which variables influence prediction.
importances = model.feature_importances_
Figure: Visualization generated to support model interpretation.
Automation
Images are saved automatically for documentation.
import os
[Link]("project_results_images", exist_ok=True)
Figure: Visualization generated to support model interpretation.
System Architecture
The system follows a structured supervised learning pipeline from data ingestion to automated reporting.
Figure: Visualization generated to support model interpretation.
Model Reliability
Reliability depends on training data quality, feature relevance, and validation strategy.
Figure: Visualization generated to support model interpretation.
Interpretability
Understanding model decisions is critical in AI projects to ensure transparency.
Figure: Visualization generated to support model interpretation.
Limitations
The model performance is dataset-dependent and may vary with new data.
Figure: Visualization generated to support model interpretation.
Ethical AI
This project is strictly academic and does not replace professional diagnosis.
Figure: Visualization generated to support model interpretation.
Future Scope
Future upgrades may include deep learning and real-time prediction systems.
Figure: Visualization generated to support model interpretation.
Learning Outcomes
This project strengthened knowledge of ML workflows, evaluation metrics, and documentation.
Figure: Visualization generated to support model interpretation.