Upload your model
Prerequisites
To upload the model you want to inspect, you need:
- A model. For example, a scikit-learn, Tensorflow, HuggingFace, catboost, PyTorch, ... Python functions
- A pandas dataframe composed of the examples you want to inspect. For example, it could be your test dataset or a dataset composed of some wrong predictions of your model
Steps to upload your model
1. Install Giskard library
In order to upload models and datasets to Giskard, you'll need to install the library giskard:
1
pip install giskard
Copied!
2. Create a new project or load an existing project
To create a new project or load an existing one, run the code below in your Python environment:
1
from giskard.giskard_client import GiskardClient
2
3
url = "http://localhost:19000" #if Giskard is installed locally (for installation, see: https://docs.giskard.ai/start/guides/installation)
4
token = "YOUR GENERATED TOKEN" #you can generate your API token in the Admin tab of the Giskard application (for installation, see: https://docs.giskard.ai/start/guides/installation)
5
client = GiskardClient(url, token)
6
7
project = client.create_project("project_key", "PROJECT_NAME", "DESCRIPTION") #Choose the arguments you want. But "project_key" should be unique and in lower case
8
#If your project is already created use project = client.get_project("existing_project_key")
Copied!
If you want to use an existing project, use
project=client.get_project("EXISTING_PROJECT_KEY")to load the existing project, then use:upload_modelto upload a new version of the model you want to inspect/testupload_datasetto upload a new dataset that you want to apply to your existing model
3. Upload a model and a dataset
Apply the upload_model_and_df to the project using the following arguments:
Argument
Description
Type
prediction_functionThe model you want to predict. It could be any Python function with the signature of
- predict_proba for classification. It returns the classification probabilities for all the classification labels
If you have preprocessing steps, wrap the whole prediction pipeline: all the preprocessing steps (categorical encoding, scaling, etc.) + ML predict_proba function.
Callable[
[pd.DataFrame], Iterable[Union[str, float, int]]
model_type- "classification" for classification model
- "regression" for regression model
str
dfA Pandas dataframe that contains some data examples that might interest you to inspect (test set, train set, production data). Some important remarks:
dfcan contain more columns than the features of the model such as the actual ground truth variable, sample_id, metadata, etc.dfshould be raw data that comes before all the preprocessing steps
Pandas dataframe
column_typesA dictionary of column names and their types (
numeric, category or text) for all columns of df. Dict[str, str]
targetThe column name in
df corresponding to the actual target variable (ground truth).str
feature_namesA list of the feature names of
prediction_function. Some important remarks:- Make sure these features are contained in
df - Make sure that
prediction_function(df[feature_names]) does not return an error message - Make sure these features have the same order as the ones used in the pipeline of
prediction_function.
List[str]
classification_labelsThe classification labels of your prediction when
prediction_task="classification". Some important remarks:- If
classification_labelsis a list of n elements, make sureprediction_functionis also returning n probabilities - Make sure the labels have the same order as the output of
prediction_function
Optional[List[str]] = None
classification_thresholdThe probability threshold in the case of a binary classification model. By default, it's equal to 0.5
Optional[float] = 0.5
model_nameThe name of the model you uploaded
Optional[str]
dataset_nameThe name of the dataset you uploaded
Optional[str]
It's better to upload
prediction_function as a function that wraps the whole prediction pipeline: all the preprocessing steps (categorical encoding, etc.) + ML prediction. This is key for a robust and interpretable inspection stage!Make sure that
prediction_function(df[feature_names]) gets executed without an error. This is the only requirement to upload a model on Giskard!Examples
1
from giskard.giskard_client import GiskardClient
2
3
client = GiskardClient(url, token)
4
5
#If you're creating your project for the first time
6
credit_scoring = client.create_project("credit_scoring", "Credit scoring project", "Predict the default probabilities of a credit demand")
7
8
#If your project is already created use
9
#project = client.get_project("credit_scoring")
10
11
credit_scoring.upload_model_and_df(
12
prediction_function=clf.predict_proba,
13
model_type='classification',
14
df=test_data,
15
column_types={
16
'credit_id':'category',
17
'credit_amount':'numeric',
18
'credit_category':'category',
19
'credit_application':'text',
20
'Is_default': 'category'
21
},
22
target = 'Is_default',
23
feature_names=['credit_amount','credit_category','credit_application'],
24
classification_labels=['Not default','Default']
25
)
Copied!
Example notebooks:
- To get started with Giskard as fast as possible we've included a demo python notebook in the platform with all the requirements on http://localhost:19000/jupyter (accessible after the Installation). Feel free to modify it to adapt it to your case!
Last modified 10d ago