import os
import json
import pandas as pd
import numpy as np
[docs]
class SummaryModel:
"""
This class summarizes the model parameters, fitness, endpoint, mode, nodes, sequence, functions, and model path.
"""
def __init__(
self, params, fitness, endpoint, mode, nodes, sequence, functions, model_path
):
"""
Initializes the SummaryModel class with the provided parameters.
Parameters:
params (pd.DataFrame): Parameters of the model.
fitness (pd.DataFrame): Fitness of the model.
endpoint (pd.DataFrame): Endpoint of the model.
mode (pd.DataFrame): Mode of the model.
nodes (pd.DataFrame): Nodes of the model.
sequence (list): Sequence of the model.
functions (list): Functions used in the model.
model_path (str): Path to the model.
"""
self.params = params
self.fitness = fitness
self.endpoint = endpoint
self.mode = mode
self.nodes = nodes
self.sequence = sequence
self.functions = functions
self.model_path = model_path
[docs]
def template(self):
"""
This function returns a formatted string that summarizes the model.
"""
return f"""
========================================================================
MODEL SUMMARY
========================================================================
🔹 Model path:
{self.model_path}
🔹 Fitness:
{self.format_dict(self.fitness)}
🔹 Endpoint:
{self.format_dict(self.endpoint)}
🔹 Mode:
{self.mode}
🔹 Nodes:
{self.format_dataframe(self.nodes)}
=========================================================================
"""
[docs]
def to_csv(self):
"""This function saves the nodes dataframe to a CSV file. If an index is provided, it appends the index to the filename for uniqueness."""
folder_name = os.path.basename(self.model_path)
# only the first part of the hash in folder_name
folder_name = folder_name.split("-")[0]
csv_filename = f"nodes_{folder_name}.csv"
csv_path = os.path.join(os.path.dirname(self.model_path), csv_filename)
self.nodes.to_csv(csv_path, index=False)
return f"Nodes saved to {csv_path}"
[docs]
def keys(self):
"""
This function returns the keys of the model.
"""
return [
"params",
"fitness",
"endpoint",
"mode",
"nodes",
"sequence",
"functions",
"model_path",
]
[docs]
def get_function(sequence, functions):
"""
Extracts the function names from the sequence based on their IDs.
The functions are from OpenCV and are used to create the model using Kartezio
(not all OpenCV functions are used). The function ID is the index of the
function in the list of functions chosen in the Kartezio package.
Parameters:
sequence (list): The sequence of the model, following the Kartezio architecture:
.. math:`[function_id, c_1, c_2, ..., c_n, p_1, p_2, ..., p_l]` where:
* function_id: Index of the function in the list of functions.
* c_i: Connections (n is the number of connections).
* p_k: Parameters (l is the number of parameters).
functions (list): The list of functions available in the Kartezio package.
Returns:
function_names (list): The list of function names corresponding to the ``function_id`` in the sequence.
"""
function_id = [sublist[0] for sublist in sequence]
function_names = [functions[i] for i in function_id]
function_names[0], function_names[1], function_names[2], function_names[-1] = (
"Hue (Input)",
"Saturation (Input)",
"Value (Input)",
"Label (Endpoint)",
)
return function_names
[docs]
def summary_model(model_folder_path):
"""
This function loads models from all subdirectories in a specified folder and creates summaries.
Parameters:
model_folder (str): Path to the folder containing the models. The function will look for subdirectories within this folder.
Returns:
summaries (list): A list of SummaryModel objects, each representing a model in the subdirectories.
"""
current = os.path.dirname(os.path.abspath(__file__))
parent = os.path.dirname(current)
model_path = os.path.join(model_folder_path, r"models")
if not os.path.exists(model_path):
raise ValueError(
f"Model path {model_path} does not exist. Please check the path."
)
subdirectories = [
d for d in os.listdir(model_path) if os.path.isdir(os.path.join(model_path, d))
]
if len(subdirectories) == 0:
raise ValueError(
f"No subdirectories found in {model_path}. Please check the path."
)
summaries = []
for subdir in subdirectories:
kartezio_model = os.path.join(model_path, subdir)
elite = os.path.join(kartezio_model, "elite.json")
if not os.path.exists(elite):
print(
f"Warning: elite.json file not found in {kartezio_model}. Skipping this subdirectory."
)
continue
with open(elite, "r") as f:
elite_data = json.load(f)
params = pd.DataFrame([elite_data["decoding"]["metadata"]])
fitness = pd.DataFrame([elite_data["individual"]["fitness"]])
endpoint = pd.DataFrame([elite_data["decoding"]["endpoint"]])
endpoint = endpoint.drop(columns=["abbv", "args", "kwargs"])
mode = pd.DataFrame([elite_data["decoding"]["mode"]])
functions = elite_data["decoding"]["functions"]
n_para = elite_data["decoding"]["metadata"]["n_para"]
n_conn = elite_data["decoding"]["metadata"]["n_conn"]
sequence = elite_data["individual"]["sequence"]
if isinstance(sequence, str):
try:
sequence = json.loads(sequence) # Convert JSON string to Python list
except json.JSONDecodeError:
raise ValueError("The 'sequence' field is not a valid JSON string.")
for i in range(len(sequence)):
if len(sequence[i]) != n_para + n_conn + 1:
raise ValueError(
f"The length of the sequence {i} is not equal to n_para."
)
conns = [sublist[1 : n_conn + 1] for sublist in sequence]
params = [sublist[n_conn + 1 :] for sublist in sequence]
function_names = get_function(sequence, functions)
nodes = pd.DataFrame(list(np.arange(1, len(sequence) + 1)), columns=[""])
nodes["Function"] = function_names
for i in range(n_conn):
nodes[f"Conn_{i+1}"] = [conn[i] for conn in conns]
for i in range(n_para):
nodes[f"Param_{i+1}"] = [param[i] for param in params]
res = SummaryModel(
params, fitness, endpoint, mode, nodes, sequence, functions, kartezio_model
)
print(res.template())
summaries.append(res)
return summaries