diff --git a/dev.py b/dev.py
deleted file mode 100644
index 1f7be3d..0000000
--- a/dev.py
+++ /dev/null
@@ -1,50 +0,0 @@
-import pandas as pd
-import numpy as np
-import scipy as sp
-import itertools
-import seaborn as sns
-
-from sklearn.model_selection import train_test_split
-from sklearn.metrics import f1_score, recall_score, roc_curve, hinge_loss
-from sklearn.ensemble import RandomForestClassifier
-from importlib import reload
-from matplotlib import pyplot as plt
-from itertools import combinations
-
-import balancers as b
-import tools
-
-
-# Reading in the data
-df = pd.read_csv('data/farm_animals.csv')
-
-# Making a single binary for shear
-shear = np.array(df.action == 'shear', dtype=np.uint8)
-shear_pred = np.array(df.pred_action == 'shear', dtype=np.uint8)
-
-# Setting up the multiple binaries
-y_bin = [np.array(y == c, dtype=np.uint8) for c in actions]
-pred_bin = [np.array(y_ == c, dtype=np.uint8) for c in actions]
-
-# Running the individual optimizations
-pbs = [b.PredictionBalancer(y_bin[i], pred_bin[i], a)
-       for i in range(len(y_bin))]
-[bal.adjust() for bal in pbs]
-bal_probs = np.array([bal.pya for bal in pbs])
-
-def conditional_matrix(pyas):
-    n_groups = pyas[0].shape[0]
-    n_outcomes = len(pyas)
-    out = np.zeros(shape=(n_outcomes, n_groups, n_groups))
-    out_range = range(n_outcomes)
-    for g in range(n_groups):
-        for col in out_range:
-            rows = list(out_range)
-            row = rows[col]
-            out[g][row][col] = pyas[g][col][1]
-            del rows[row]
-            for r in rows:
-                out[g][r][col] = pyas[g][col][0]
-    return out
-        
-