issues in ISRO SPACE mining .py file

[Sakeebhasan123456]

@Devanik21 i analyze the mining python file and see lots of abnormalities ,i would to work on it

there is list of problem s that i can solve

Data Leakage:** Pre-trained models may have been trained using the full dataset, leading to inflated performance.

Overfitting:** Models show unrealistically high accuracy, suggesting they are overfitting to the training data.

Imbalanced Dataset:** The target variable might be imbalanced, which could bias the model’s predictions.

I### nadequate Evaluation:** Overreliance on accuracy without considering metrics like ROC AUC or F1-score.

No Cross-Validation:** Cross-validation is missing, which could lead to overestimating model performance.

Missing Feature Scaling:** Numerical features are not scaled, which may affect the performance of some models.

Categorical Encoding:** The categorical feature 'Celestial Body' is not properly encoded.

Unused Imports:** Libraries like TensorFlow and Keras are imported but not used.

Inconsistent Random Seeds:** Random seeds are not consistently set, making results non-reproducible.

No Hyperparameter Tuning:** Models are used with default parameters without sufficient tuning.

[Devanik21]

@Devanik21 i analyze the mining python file and see lots of abnormalities ,i would to work on it

there is list of problem s that i can solve

Data Leakage:** Pre-trained models may have been trained using the full dataset, leading to inflated performance.

Overfitting:** Models show unrealistically high accuracy, suggesting they are overfitting to the training data.

Imbalanced Dataset:** The target variable might be imbalanced, which could bias the model’s predictions.

I### nadequate Evaluation:** Overreliance on accuracy without considering metrics like ROC AUC or F1-score.

No Cross-Validation:** Cross-validation is missing, which could lead to overestimating model performance.

Missing Feature Scaling:** Numerical features are not scaled, which may affect the performance of some models.

Categorical Encoding:** The categorical feature 'Celestial Body' is not properly encoded.

Unused Imports:** Libraries like TensorFlow and Keras are imported but not used.

Inconsistent Random Seeds:** Random seeds are not consistently set, making results non-reproducible.

No Hyperparameter Tuning:** Models are used with default parameters without sufficient tuning.

Thanks .
well said, plz try with all of them with overfitting issue, cross validation , hyper parameter tuning in priority.

[JahnaviDhanaSri]

Hi @Devanik21

I would like to take on the task of resolving the identified issues in the ISRO SPACE mining .py file.
I will analyze the model's training process to implement regularization techniques and potentially adjust the model architecture to reduce overfitting.
I intend to introduce k-fold cross-validation to ensure more reliable model evaluation and performance metrics.
I will utilize techniques like Grid Search or Random Search to optimize hyperparameters and enhance the model's predictive accuracy.

[Devanik21]

Hi @Devanik21

I would like to take on the task of resolving the identified issues in the ISRO SPACE mining .py file. I will analyze the model's training process to implement regularization techniques and potentially adjust the model architecture to reduce overfitting. I intend to introduce k-fold cross-validation to ensure more reliable model evaluation and performance metrics. I will utilize techniques like Grid Search or Random Search to optimize hyperparameters and enhance the model's predictive accuracy.

hello , are you working on it?

[JahnaviDhanaSri]

Yeah I am working on it