diff --git a/survivors/ensemble/clever_boosting_old.py b/survivors/ensemble/clever_boosting_old.py
deleted file mode 100644
index 9c6c3d6..0000000
--- a/survivors/ensemble/clever_boosting_old.py
+++ /dev/null
@@ -1,140 +0,0 @@
-import numpy as np
-
-from .. import metrics as metr
-from ..tree import CRAID
-from .. import constants as cnt
-from .boosting import BoostingCRAID
-import matplotlib.pyplot as plt
-
-
-class IBSCleverBoostingCRAID(BoostingCRAID):
-    def __init__(self, **kwargs):
-        super().__init__(**kwargs)
-        self.name = "IBSCleverBoostingCRAID"
-
-    def fit(self, X, y):
-        self.features = X.columns
-        X = X.reset_index(drop=True)
-        X[cnt.CENS_NAME] = y[cnt.CENS_NAME].astype(np.int32)
-        X[cnt.TIME_NAME] = y[cnt.TIME_NAME].astype(np.float32)
-
-        self.X_train = X
-        self.X_train["ind_start"] = self.X_train.index
-        self.y_train = y
-
-        self.weights = np.ones(self.X_train.shape[0], dtype=float)
-        self.bettas = []
-        self.l_ibs = []
-        self.l_weights = []
-        self.update_params()
-
-        for i in range(self.n_estimators):
-            x_sub = self.X_train.sample(n=self.size_sample,
-                                        # weights=self.weights,
-                                        replace=self.bootstrap, random_state=i)
-
-            x_oob = self.X_train.loc[self.X_train.index.difference(x_sub.index), :]
-            # print(f"UNIQUE ({i}):{np.unique(x_sub.index).shape[0]}, DIST:", np.bincount(x_sub["cens"]))
-            x_sub = x_sub.reset_index(drop=True)
-            X_sub_tr, y_sub_tr = cnt.pd_to_xy(x_sub)
-            if self.weighted_tree:
-                X_sub_tr["weights_obs"] = self.weights[x_sub['ind_start']]  # self.weights
-
-            model = CRAID(features=self.features, random_state=i, **self.tree_kwargs)
-            model.fit(X_sub_tr, y_sub_tr)
-
-            wei_i, betta_i = self.count_model_weights(model, X_sub_tr, y_sub_tr)
-            self.add_model(model, x_oob, wei_i, betta_i)
-            self.update_weight(x_sub['ind_start'], wei_i)
-
-    def predict(self, x_test, aggreg=True, **kwargs):
-        res = []
-        weights = []
-        for i in range(len(self.models)):
-            res.append(self.models[i].predict(x_test, **kwargs))
-
-        res = np.array(res)
-        weights = None  # np.vstack(weights).T
-        if aggreg:
-            res = self.get_aggreg(res, weights)
-        return res
-
-    def predict_at_times(self, x_test, bins, aggreg=True, mode="surv"):
-        res = []
-        weights = []
-        for i in range(len(self.models)):
-            res.append(self.models[i].predict_at_times(x_test, bins=bins,
-                                                       mode=mode))
-
-        res = np.array(res)
-        weights = None  # np.vstack(weights).T
-        if aggreg:
-            res = self.get_aggreg(res, weights)
-            if mode == "surv":
-                res[:, -1] = 0
-                res[:, 0] = 1
-        return res
-
-    def count_model_weights(self, model, X_sub, y_sub):
-        if self.all_weight:
-            X_sub = self.X_train
-            y_sub = self.y_train
-        pred_sf = model.predict_at_times(X_sub, bins=self.bins, mode="surv")
-
-        # PRED WEI!!!
-        ibs_sf = metr.ibs_WW(self.y_train, y_sub, pred_sf, self.bins, axis=0)
-
-        if len(self.bettas) > 0:
-            pred_ens = self.predict_at_times(X_sub, bins=self.bins, mode="surv")
-            ibs_ens = metr.ibs_WW(self.y_train, y_sub, pred_ens, self.bins)
-            betta = ibs_ens / np.mean(ibs_sf)
-        else:
-            betta = 1
-
-        wei = ibs_sf
-        return wei, abs(betta)
-
-    def update_weight(self, index, wei_i):
-        # PRED WEI!!!
-        if len(self.models) > 1:
-            self.weights = self.weights + (self.bettas[-1] ** 2) * wei_i
-            self.weights /= (1 + self.bettas[-1]) ** 2
-            self.bettas = list(np.array(self.bettas) / np.sum(self.bettas))
-        else:
-            self.weights = wei_i
-
-    def get_aggreg(self, x, wei=None):
-        if self.aggreg_func == 'median':
-            return np.median(x, axis=0)
-        elif self.aggreg_func == "wei":
-            if wei is None:
-                wei = np.array(self.bettas)
-            wei = wei / np.sum(wei)
-            return np.sum((x.T * wei).T, axis=0)
-        elif self.aggreg_func == "argmean":
-            wei = np.where(np.argsort(np.argsort(wei, axis=1), axis=1) > len(self.bettas) // 2, 1, 0)
-            wei = wei / np.sum(wei, axis=1).reshape(-1, 1)
-            return np.sum((x.T * wei).T, axis=0)
-        elif self.aggreg_func == "argwei":
-            wei = np.where(np.argsort(np.argsort(wei, axis=1), axis=1) > len(self.bettas) // 2,
-                           1 / np.array(self.bettas), 0)
-            wei = wei / np.sum(wei, axis=1).reshape(-1, 1)
-            return np.sum((x.T * wei).T, axis=0)
-        return np.mean(x, axis=0)
-
-    def plot_curve(self, X_tmp, y_tmp, bins, label="", metric="ibs"):
-        res = []
-        metr_vals = []
-        for i in range(len(self.models)):
-            res.append(self.models[i].predict_at_times(X_tmp, bins=bins, mode="surv"))
-
-            res_all = np.array(res)
-            res_all = self.get_aggreg(res_all, np.array(self.bettas)[:i + 1])
-            res_all[:, -1] = 0
-            res_all[:, 0] = 1
-            if metric == "ibs":
-                metr_vals.append(metr.ibs_WW(self.y_train, y_tmp, res_all, bins))
-            else:
-                metr_vals.append(metr.auprc(self.y_train, y_tmp, res_all, bins))
-        plt.plot(range(len(self.models)), metr_vals, label=label)
-