Installation#
Prerequisite#
Modeva is developed for Python >= 3.9 and Python <= 3.12, with following dependencies:
Dev Ops
Dependency |
Minimum Version |
Maximum Version |
|---|---|---|
mlflow |
1.28.0 |
|
supabase |
Dashboard & Visualization
Dependency |
Minimum Version |
Maximum Version |
|---|---|---|
ipywidgets |
7.7.0 |
|
ipython |
7.12.0 |
|
ipyvuetify |
1.8.10 |
|
notebook |
7.0.0 |
|
mocharts |
1.0.0 |
Basic Computing
Dependency |
Minimum Version |
Maximum Version |
|---|---|---|
numpy |
1.21.4 |
1.26.4 |
scipy |
1.5.3 |
|
pandas |
1.3.5 |
Machine Learning
Dependency |
Minimum Version |
Maximum Version |
|---|---|---|
scikit-learn |
1.3.0 |
1.6.1 |
scikit-learn-extra |
||
xgboost |
1.4.2 |
|
catboost |
||
lightgbm |
||
torch |
1.11.0 |
|
momentchi2 |
0.1.8 |
|
umap-learn |
||
pyswarms |
||
lime |
0.2.0.1 |
|
shap |
0.39.0 |
Miscellaneous
Dependency |
Minimum Version |
Maximum Version |
|---|---|---|
dill |
||
tqdm |
||
httpx |
||
python-dateutil |
Installation#
You can install the package by pip:
pip install modeva -U
Trouble Shooting#
If you have installed old version of Modeva, you may encounter issues due to conflict of mlflow and its caches.
Command line for updating mlflow package.
pip install mlflow -U
Python codes to clean up existing mlflow caches.
from modeva.utils.mlflow import clear_mlflow_home
clear_mlflow_home()