Crop_Yield_Prediction_Presentation.pptx Sample contet

Data Anlytics in Farming

1. TITLE : Crop Yield Prediction using ML & Data Visualization Monish D N | Project Presentation | 2025

2. Introduction Agriculture is the backbone of India’s economy. Crop yield prediction is vital for farmers and policymakers and researchers Objective: Predict yields using soil, climate, and crop features.

3. Objectives Predict crop yield for different crops, states, and seasons. Compare ML models (Linear Regression, Random Forest, Gradient Boosting). Provide interactive dashboards in Power BI.

4. Dataset Description Source: Kaggle & Indian agricultural datasets. Features: Crop, Season, State, Year, Area, Production. Climate: Rainfall, Fertilizer, Pesticide usage. Soil: pH, Nitrogen, Phosphorus, Potassium. Target: Yield (Production/Area).

5. Data Preprocessing Handled missing values. Encoded categorical variables (Crop, State, Season, Soil Type). Normalized numeric features. Split into Train/Test sets (80/20).

6. Exploratory Data Analysis Yield and Rainfall distribution (histograms). Rainfall vs Yield (scatterplots). Seasonal yield variation (bar charts). State-wise yield performance by Soil Type (bar chart). Correlation heatmap.

7. Fig 1 : Top 10 States by Average Yield Fig 2 : Distribution of Annual Rainfall Fig 3 : Average Yield by States Fig 4 : Average Yield by Season

8. Fig 5 : Heatmap Fig 6 : Distribution of Crop Yield Fig 7 : State Wise Yield by Soil Type

9. Machine Learning Models Linear Regression – baseline model – linear relationships . Random Forest – ensemble model - non-linear relationships. Gradient Boosting – optimization technique – sequential learner from errors .

10. Model Evaluation Metrics RMSE – Root Mean Squared Error. MAE – Mean Absolute Error. R² – Variance explained by model. Accuracy – R 2 %

11. Results Summary Linear Regression: Accuracy ~4 8.8 %. Random Forest: Accuracy ~ 88 %. Gradient Boosting: Accuracy ~94.7%. Gradient Boosting performed best.

12. DASHBOARD IMPLEMENTATION Saved actual vs predicted yields into CSV. Imported results into Power BI dashboard. Turned data into insights and presented insights in clear and actionable format.

13. Fig 8 : Dashboard implementation with Power BI

14. Conclusion Gradient Boosting achieved 94.7% accuracy. Dataset is highly skewed - some extreme yields cause high RMSE Soil features improved predictions. Visualization revealed yield hotspots and seasonal trends. Supports farmers, policymakers, and researchers.

15. Real-World Impact Helps farmers optimize crop planning and inputs. Supports government in subsidy & policy decisions. Enables researchers to study crop-soil suitability. Minimizes risks from unpredictable yields.

16. THANK YOU