Skip to content

Applying Your Model

This guide demonstrates how to apply a trained model to generate predictions. You can use either the Python API for programmatic control or the command-line tools for a code-free approach.

Method 1: Python API

This method offers model application process within a Python script.

Prerequisites

Before you begin, make sure you have:

  1. Installed the Python API for MES Infrastructure.
  2. Loaded your data repository by following the ETL Tutorial.
  3. Created your testing dataset as outlined in the Create Samples Tutorial.

Inference Script

The following script shows how to perform model inference from start to finish.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
import med

# --- Configuration ---
# Path to your data repository
rep_path = ''
# Path to your trained MedModel file
model_file = ''
# Path to your samples file. Alternatively, load from a DataFrame
# using: samples.from_df(dataframe_object)
samples_file = ''

# --- Initialization ---
print("Initializing repository...")
rep = med.PidRepository()
# Initialize for the first processing step
rep.init(rep_path)

# --- Load Model ---
print("Loading model...")
model = med.Model()
model.read_from_file(model_file)
model.fit_for_repository(rep)
# Get the list of signals required by the model
signalNamesSet = model.get_required_signal_names()

# --- Load Samples ---
print("Loading samples...")
samples = med.Samples()
samples.read_from_file(samples_file)
# Get the IDs from the samples to fetch relevant data
ids = samples.get_ids()

# --- Read Data ---
print("Reading data from repository...")
# Read only the necessary data for the specified IDs and signals
rep.read_all(rep_path, ids, signalNamesSet)

# --- Apply Model ---
print("Applying model to samples...")
model.apply(rep, samples)

# --- Save Results ---
# The 'samples' object now contains the model scores
# Convert to a DataFrame for further analysis
df = samples.to_df()
# Save the results to a CSV file
df.to_csv('output_file.csv', index=False)
# Or save as a samples file
samples.write_to_file('output_samples.tsv')

# The feature matrix is available via: model.features.to_df()
print("Inference complete. Results saved.")

[!NOTE] To create feature matrix of the model, apply the model and in the end use model.features.to_df() to extract the feature matrix into dataframe.

Method 2: Command-Line Tools

If you prefer to work from the command line, MES offers tools to apply models without writing any Python code.

Prerequisites

First, install the MES Tools for Training and Testing Models.

Prediction Tools

Use the following tool to generate predictions or extract the full feature matrix from your model's pipeline: