📖 Student_Final_Prediction -- ML Workflow

1: Problem definition

student final prediction = hadafka wa in la sii sadaalin karo final ka ardeyga si maamulayaasha ardeyda liita muraajico u sameyan kuwa top ku jira la abaal mariyo, kuwa dhexe na la dhiiri galiyo, wana in xog cusub aqbali karaa, project gu waana Supervised qeybta Classfication.

2: Data collection

Dataset ku waa 1k+ ardey ah, 9 columns 8 ka mid ah waa feutures 1 waa label

3: Data preprocessing

waxan ka saaray khaladaadka, meelaha banan oo aan buuxiyey, duplicate, iqr_function, feauture engineering, One Hot Encoding, save the scaler and training features, saved dataset

4: check images

single person check

upload file

before

after

5: Model selection

• Ka fiiri dhowr algorithm oo ku habboon haddii final score tiro yahay: Linear Regression, Random Forest Regressor, XGBoost) si aan u helo pass/fail.
• Samee train/test split (e.g., 80/20).

6: Model evaluation

• Train/test split, scaler
• random forest, logistic regression, decition tree, xgboost
• single row sanit check
• custom inputs
• Metrics:
  • Classification → Accuracy, Precision, Recall, F1-score
• saved models

---

7: Deployment & usage

• API (Flask) si xog cusub loo galiyo loona helo prediction
• frontend HTML + CSS + JS si macallimiinta ama maamulka si fudud uga arkaan

8: Monitoring & improvement

• Si joogto ah u cusbooneysii modelka marka xog cusub timaado
• Ka war hay haddii performance uu hoos u dhaco model drift
• kusoo daray feutures sida in excel lasoo galin karo

9: Who uses it?

managers to decide whether students who fail the final exam at a school, university or college

10: 💡 How is it used?

Fadlan excel file kaaga columns kaan wa in ku jiran Attendance Assignments Quiz Midterm

weliba sida ay u qoran yihiin, lakiin wax kale waad kuso darsan karta sida name,age,gender,id

kadibna soo download gareyso list kaliya ah fail examka dhici raba finalka si aad uga hortagto oo aad murajico ugu sameyso

11: 💡scores/ dhibcaha

1: Attendance = 0 ilaa 5 2: Assignments = 0 ilaa 10

3: Quiz = 0 ilaa 5 4: Midterm = 0 ilaa 20

• totalku waa 40 maadama u finalka yahay 60

---

12: 🤖 Models

waxan ku toobaray:

• logistic Regression
• Random Forest
• decition tree
• XGBoost

Saved models:

• modeles/lr_model.joblib
• modeles/rf_model.joblib
• modeles/ds_model.joblib
• modeles/rf_model.joblib

---

13: Working API deployment

---

14: 🚀 Usage

Run via API

Start Flask API:

Send POST request

http://localhost:8000/predict?model=lr http://localhost:8000/predict?model=rf http://localhost:8000/predict?model=df http://localhost:8000/predict?model=xgb

---

Example JSON input:

{ "Attendance": 4 "Assignments": 8
"Quiz": 5 "Midterm": 12 }

Example response:

{ "model": "logistic regression", "Total_score": 29, "prediction_result": fail }

{ "model": "XGBoost", "Total_score": 29, "prediction_result": pass }

---

🏠 House Rent Prediction

Welcome! Machine Learning Project to Predict House Prices

📖 Description

project gan waa predict house rent based on features such as square, number of bedrooms, bathrooms, year built, and location

waxan ku toobaray Linear Regression and Random Forest models kadibna waxan ku xiray frontend HTML, CSS, JS

tijabo:

Linear Regression(lr)

Random Forest(rf)

- 📊 Dataset

• data waa clean waana la preprocessed gareyey
• ku saved gareyey: dataset/house_l0000_Clean_dataset.csv

---

🧾 Features

Main features used for prediction:

Feature	Description
Size_sqft	cabirka ku fadhiyo
Bedrooms	tirada qolalka
Bathrooms	tirada suuli yada
YearBuilt	sanadka la dhisay
Location	meesha u ku yaal (City, Suburb, Rural)
HouseAge	inta sano uu jiro gurigaas ma cusayb mise duug
Rooms_per_1000sqft	celceliska qolalka
Size_per_Bedroom	cabirka qolalka
Is_City	magaalo ma ku yaal ? (1=City)

🤖 Models

waxan ku toobaray:

• Linear Regression
• Random Forest Regressor

Saved models:

• modeles/lr_model.joblib
• modeles/rf_model.joblib

---

🧹 Preprocessing Steps

1. Handling missing values (sixida waxa maqan ):
   • Size_sqft → median
   • Bedrooms → mode
   • Location → mode
2. Remove duplicates(ka saar waxa 2 jeer so labtay)
3. IQR capping for Price and Size_sqft
4. One-hot encoding for Location
5. Feature engineering: HouseAge, Rooms_per_1000sqft, Size_per_Bedroom, Is_City
6. Feature scaling: StandardScaler applied to numerical features

---

📈 Evaluation

• RMSE
• MAE
• R² Score

---

🚀 Usage

Run via API

Start Flask API:

Send POST request

http://localhost:8000/predict?model=lr http://localhost:8000/predict?model=rf

Example JSON input:

{ "Size_sqft": 2000, "Bedrooms": 3, "Bathrooms": 2, "YearBuilt": 2010, "Location": "City" }

Example response:

{ "model": "linear_regression", "prediction": 230000.0 }

{ "model": "random_forest", "prediction": 250000.0 }