Upload folder using huggingface_hub
Browse files- .gitattributes +1 -0
- data/metadata.json +84 -0
- images/confusion_matrix.png +0 -0
- images/shap_feature_importance_plot.png +0 -0
- model/stroke_risk_prediction/MLmodel +48 -0
- model/stroke_risk_prediction/conda.yaml +14 -0
- model/stroke_risk_prediction/input_example.json +0 -0
- model/stroke_risk_prediction/model.cb +3 -0
- model/stroke_risk_prediction/python_env.yaml +7 -0
- model/stroke_risk_prediction/registered_model_meta +2 -0
- model/stroke_risk_prediction/requirements.txt +7 -0
- model/stroke_risk_prediction/serving_input_example.json +0 -0
- preprocessor/preprocessor.joblib +3 -0
.gitattributes
CHANGED
|
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
model/stroke_risk_prediction/model.cb filter=lfs diff=lfs merge=lfs -text
|
data/metadata.json
ADDED
|
@@ -0,0 +1,84 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"random_seed": 14,
|
| 3 |
+
"test_size": 0.2,
|
| 4 |
+
"target_feature": "stroke",
|
| 5 |
+
"original_features": [
|
| 6 |
+
"gender",
|
| 7 |
+
"age",
|
| 8 |
+
"hypertension",
|
| 9 |
+
"heart_disease",
|
| 10 |
+
"ever_married",
|
| 11 |
+
"work_type",
|
| 12 |
+
"Residence_type",
|
| 13 |
+
"avg_glucose_level",
|
| 14 |
+
"bmi",
|
| 15 |
+
"smoking_status"
|
| 16 |
+
],
|
| 17 |
+
"binary_encoding": {
|
| 18 |
+
"columns": [
|
| 19 |
+
"ever_married"
|
| 20 |
+
],
|
| 21 |
+
"mapping": {
|
| 22 |
+
"No": 0,
|
| 23 |
+
"Yes": 1
|
| 24 |
+
}
|
| 25 |
+
},
|
| 26 |
+
"imputation": {
|
| 27 |
+
"method": "SimpleImputer",
|
| 28 |
+
"strategy": "median",
|
| 29 |
+
"columns": [
|
| 30 |
+
"bmi"
|
| 31 |
+
]
|
| 32 |
+
},
|
| 33 |
+
"scaling": {
|
| 34 |
+
"method": "StandardScaler",
|
| 35 |
+
"columns": [
|
| 36 |
+
"avg_glucose_level",
|
| 37 |
+
"bmi"
|
| 38 |
+
]
|
| 39 |
+
},
|
| 40 |
+
"encoding": {
|
| 41 |
+
"method": "OneHotEncoder",
|
| 42 |
+
"columns": [
|
| 43 |
+
"gender",
|
| 44 |
+
"work_type",
|
| 45 |
+
"residence_type",
|
| 46 |
+
"smoking_status"
|
| 47 |
+
],
|
| 48 |
+
"verbose_feature_names": false
|
| 49 |
+
},
|
| 50 |
+
"processed_features": [
|
| 51 |
+
"stroke",
|
| 52 |
+
"ever_married",
|
| 53 |
+
"heart_disease",
|
| 54 |
+
"hypertension",
|
| 55 |
+
"age",
|
| 56 |
+
"smoking_status_smokes",
|
| 57 |
+
"smoking_status_never_smoked",
|
| 58 |
+
"smoking_status_formerly_smoked",
|
| 59 |
+
"smoking_status_unknown",
|
| 60 |
+
"residence_type_urban",
|
| 61 |
+
"residence_type_rural",
|
| 62 |
+
"work_type_children",
|
| 63 |
+
"work_type_self-employed",
|
| 64 |
+
"work_type_private",
|
| 65 |
+
"work_type_never_worked",
|
| 66 |
+
"work_type_govt_job",
|
| 67 |
+
"gender_other",
|
| 68 |
+
"gender_male",
|
| 69 |
+
"gender_female",
|
| 70 |
+
"bmi",
|
| 71 |
+
"avg_glucose_level"
|
| 72 |
+
],
|
| 73 |
+
"artifacts_dir": "data/artifacts",
|
| 74 |
+
"processed_data_dir": "data/processed",
|
| 75 |
+
"train_shape": [
|
| 76 |
+
4088,
|
| 77 |
+
21
|
| 78 |
+
],
|
| 79 |
+
"test_shape": [
|
| 80 |
+
1022,
|
| 81 |
+
21
|
| 82 |
+
],
|
| 83 |
+
"timestamp": "2026-05-16T16:59:22.957524"
|
| 84 |
+
}
|
images/confusion_matrix.png
ADDED
|
images/shap_feature_importance_plot.png
ADDED
|
model/stroke_risk_prediction/MLmodel
ADDED
|
@@ -0,0 +1,48 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
artifact_path: /media/vini/DATA/Git/ml_full_project/mlruns/3/models/m-39de93ac07234ce09b688ba8df87c48d/artifacts
|
| 2 |
+
flavors:
|
| 3 |
+
catboost:
|
| 4 |
+
catboost_version: 1.2.10
|
| 5 |
+
code: null
|
| 6 |
+
data: model.cb
|
| 7 |
+
model_type: CatBoostClassifier
|
| 8 |
+
save_format: cbm
|
| 9 |
+
python_function:
|
| 10 |
+
data: model.cb
|
| 11 |
+
env:
|
| 12 |
+
conda: conda.yaml
|
| 13 |
+
virtualenv: python_env.yaml
|
| 14 |
+
loader_module: mlflow.catboost
|
| 15 |
+
python_version: 3.12.11
|
| 16 |
+
is_signature_from_type_hint: false
|
| 17 |
+
mlflow_version: 3.10.1
|
| 18 |
+
model_id: m-39de93ac07234ce09b688ba8df87c48d
|
| 19 |
+
model_size_bytes: 2012951
|
| 20 |
+
model_uuid: m-39de93ac07234ce09b688ba8df87c48d
|
| 21 |
+
prompts: null
|
| 22 |
+
run_id: 5c4cf23fc7c94efa8b9ffcbf2265c7a1
|
| 23 |
+
saved_input_example_info:
|
| 24 |
+
artifact_path: input_example.json
|
| 25 |
+
pandas_orient: split
|
| 26 |
+
serving_input_path: serving_input_example.json
|
| 27 |
+
type: dataframe
|
| 28 |
+
signature:
|
| 29 |
+
inputs: '[{"type": "double", "name": "avg_glucose_level", "required": true}, {"type":
|
| 30 |
+
"double", "name": "bmi", "required": true}, {"type": "double", "name": "gender_female",
|
| 31 |
+
"required": true}, {"type": "double", "name": "gender_male", "required": true},
|
| 32 |
+
{"type": "double", "name": "gender_other", "required": true}, {"type": "double",
|
| 33 |
+
"name": "work_type_govt_job", "required": true}, {"type": "double", "name": "work_type_never_worked",
|
| 34 |
+
"required": true}, {"type": "double", "name": "work_type_private", "required":
|
| 35 |
+
true}, {"type": "double", "name": "work_type_self-employed", "required": true},
|
| 36 |
+
{"type": "double", "name": "work_type_children", "required": true}, {"type": "double",
|
| 37 |
+
"name": "residence_type_rural", "required": true}, {"type": "double", "name":
|
| 38 |
+
"residence_type_urban", "required": true}, {"type": "double", "name": "smoking_status_unknown",
|
| 39 |
+
"required": true}, {"type": "double", "name": "smoking_status_formerly_smoked",
|
| 40 |
+
"required": true}, {"type": "double", "name": "smoking_status_never_smoked", "required":
|
| 41 |
+
true}, {"type": "double", "name": "smoking_status_smokes", "required": true},
|
| 42 |
+
{"type": "double", "name": "age", "required": true}, {"type": "double", "name":
|
| 43 |
+
"hypertension", "required": true}, {"type": "double", "name": "heart_disease",
|
| 44 |
+
"required": true}, {"type": "double", "name": "ever_married", "required": true}]'
|
| 45 |
+
outputs: '[{"type": "long", "required": true}]'
|
| 46 |
+
params: null
|
| 47 |
+
type_hint_from_example: false
|
| 48 |
+
utc_time_created: '2026-05-16 16:06:46.813357'
|
model/stroke_risk_prediction/conda.yaml
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
channels:
|
| 2 |
+
- conda-forge
|
| 3 |
+
dependencies:
|
| 4 |
+
- python=3.12.11
|
| 5 |
+
- pip<=26.0.1
|
| 6 |
+
- pip:
|
| 7 |
+
- mlflow==3.10.1
|
| 8 |
+
- catboost==1.2.10
|
| 9 |
+
- colorama==0.4.6
|
| 10 |
+
- ipython==9.12.0
|
| 11 |
+
- numpy==2.4.4
|
| 12 |
+
- pandas==2.3.3
|
| 13 |
+
- scipy==1.17.1
|
| 14 |
+
name: mlflow-env
|
model/stroke_risk_prediction/input_example.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
model/stroke_risk_prediction/model.cb
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8f09d323dc3bccd3276d6fe73560f824dec1ffc44eb1112ee0cfbc033b71296f
|
| 3 |
+
size 200392
|
model/stroke_risk_prediction/python_env.yaml
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
python: 3.12.11
|
| 2 |
+
build_dependencies:
|
| 3 |
+
- pip==26.0.1
|
| 4 |
+
- setuptools==82.0.1
|
| 5 |
+
- wheel
|
| 6 |
+
dependencies:
|
| 7 |
+
- -r requirements.txt
|
model/stroke_risk_prediction/registered_model_meta
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_name: stroke_risk_prediction
|
| 2 |
+
model_version: '1'
|
model/stroke_risk_prediction/requirements.txt
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
mlflow==3.10.1
|
| 2 |
+
catboost==1.2.10
|
| 3 |
+
colorama==0.4.6
|
| 4 |
+
ipython==9.12.0
|
| 5 |
+
numpy==2.4.4
|
| 6 |
+
pandas==2.3.3
|
| 7 |
+
scipy==1.17.1
|
model/stroke_risk_prediction/serving_input_example.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
preprocessor/preprocessor.joblib
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ef69e32a290c075d3220709ad81cd7375b8654c30fc6c80f7c72fd1c5f153d3b
|
| 3 |
+
size 4063
|