optimize_spinemet-selection.py

import pandas as pd
import numpy as np
import tensorflow as tf
tf.compat.v1.enable_eager_execution()
tf.compat.v1.random.set_random_seed(42)

from tensorflow_probability import distributions as tfd
from tensorflow.keras.layers import Input, Dense, Activation, Concatenate, BatchNormalization, Dropout
from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint, ReduceLROnPlateau

from sksurv.metrics import concordance_index_ipcw, integrated_brier_score, cumulative_dynamic_auc
from sksurv.util import Surv
from sklearn.metrics import mean_squared_error
from sklearn.model_selection import train_test_split

from model import MDN
from utils import NLLLoss,CensoredNLLLoss,AlternativeNLLLoss
from utils import concordance_index_ipcw_scorer,integrated_brier_scorer,cumulative_dynamic_auc_scorer,root_mean_squared_error_scorer

import optuna

from optimizer import Optimizer
from sklearn.preprocessing import MinMaxScaler

import json

from sksurv.datasets import load_veterans_lung_cancer
from sksurv.column import encode_categorical

name = "MDN_spinemet_selection_data"

# Load and Preprocess
df = pd.read_csv("Data/spinemet_selection.tsv", sep = "\t")
drop_list = ["nmstpc","categories"] # These are predictions...
df = df.drop(drop_list,axis=1)
time_scaler = MinMaxScaler()
df["survival"]= time_scaler.fit_transform(df["survival"].to_numpy().reshape(-1, 1))
t = np.float32(df["survival"].to_numpy())
delta = df["census"].to_numpy().astype(np.float32)
from preprocessing import Preprocessor
num_feats = ["primer_tumor","age","mobility","metastasis","protein"]        
pp = Preprocessor(cat_feat_strat="mode",num_feat_strat="knn",scaling_strategy="minmax",remaining="drop")
df = pp.fit_transform(df, cat_feats=[], num_feats=num_feats)
X = df.copy()
x_size = len(X.columns)
X = np.float32(X.to_numpy())
y = np.stack([t,delta],axis = 1)

# Run Optimizer
print("Running Optimizer with LogRank binary scoring")
opt = Optimizer(X,y,name,num_epochs=200,batch_size=128,use_kfold=True,use_logrank=True,
                kernel_list = ["Exponential","Gumbel","Normal","Logistic","LogLogistic"])
best_val,best_params = opt(1000)
best_params['UnoC_LR']=best_val
print(best_params)
with open(f"Logs/{name}.json", "w") as write_file:
    json.dump(best_params, write_file)
    
print("Running Optimizer normaly")
opt = Optimizer(X,y,name,num_epochs=200,batch_size=128,use_kfold=True,use_logrank=False,
                kernel_list = ["Exponential","Gumbel","Normal","Logistic","LogLogistic"])
best_val,best_params = opt(1000)
best_params['UnoC']=best_val
print(best_params)
with open(f"Logs/{name}_no_logrank.json", "w") as write_file:
    json.dump(best_params, write_file)