Direct labeling: continuous creation of training dataset

Course Q&A Machine Learning Engineering for Production(MLOps) Machine Learning Data Lifecycle in Production
1

If you look at the image below, I am confused about what he means by ‘inference requests’ in the first box all the way to the left. Can someone explain what he means by this?

image
image1006×435 43.2 KB

2

Hi @Nathan_Angell,

When you deploy your machine learning model in a production environment, people utilize it to make predictions. This process of using the model to make predictions is commonly referred to as an “inference request.” Inference requests can come in various formats depending on the specific deployment scenario and environment.

For instance, you might receive an inference request in JSON format. If your model is predicting something like student performance, the request could include input features like “Age” or “Gender,” among others. These features are typically passed as part of the JSON-formatted request to the deployed model.

To illustrate further, imagine a scenario where a user wants to predict a student’s performance. The inference request, formatted in JSON, might look like this:

{
   "Age": 18,
   "Gender": "Male",
   "Previous_Grades": [85, 92, 78],
   "Study_Hours": 4,
   "Extracurricular_Activities": ["Chess Club", "Debate Team"]
}

In this JSON request, the model receives information about the student’s age, gender, previous grades, study hours, and extracurricular activities. The model processes this data to generate a prediction.

I hope this clarifies the concept of an inference request for you. Please feel free to ask if you need further clarification or have more questions.

Cheers!
Jamal

3

If we decided to predict a student’s performance the second box, located in the image I sent, would represent what the model spits out when it tries to predict the student’s performance. Is that correct? @Jamal022

4

Hey @Nathan_Angell,

Sorry but do you mean the label or classification output of the model by “Spitting” ? :sweat_smile:

in case that’s what you mean then yes you are correct that would typically represent the output or prediction generated by the machine learning model when it tries to predict a student’s performance based on the information provided in the inference request.

1 Like