return conver_to_probs(votes)
if __name__=="__main__":
#net_file = 'res/9503-jaccard-votingnets.pickle'
#test_file ='res/10003-test.csv'
net_file = 'res/9003-tc-jaccard-votingnets.pickle'
test_file = 'res/test-tc-1000.npy'
#test = pd.read_csv(test_file, index_col=0)
test = np.load(test_file)
with open(net_file, 'rb') as f:
nets = pickle.load(f)
m = get_data()
distance = lambda x,y: jaccard(x,y, m['nf'])
classifier = lambda x: lcl(x, m['ml'])
refs = dict()
mers = dict()
#test_a = list(test.index)
test_a = list(test)
for gamma, [mer, ref] in tqdm(nets.items()):
mers[gamma] = get_predictions_score(net=mer, distance=distance, test=test_a)
refs[gamma] = get_predictions_score(net=ref, distance=distance, test=test_a)
with open(f"/tmp/preds-{gamma}-mers.pickle", 'wb+') as f:
pickle.dump(mers[gamma], f)
with open(f"/tmp/preds-{gamma}-refs.pickle", 'wb+') as f:
pickle.dump(refs[gamma], f)
def distance(x, y, arg):
return jaccard(x, y, arg)
def mjaccard(x, y, nafs):
if len(nafs[x]) == 0 == len(nafs[y]):
return 1
return jaccard(x, y, nafs)
sizes = {g:len(n.keys()) for g, n in nets.items()}
perfs_di = pd.DataFrame.from_dict(res, orient='index', columns=['fp', 'fn'])
szs_di = pd.DataFrame.from_dict(sizes, orient='index', columns=['sizes'])
gj_euc = perfs_di.join(szs_di)
part_size = len(train)
gj_euc['compression'] = gj_euc.sizes/part_size
gj_euc.to_csv(f"res/euc{part_size}.csv")
if __name__ == "__main__":
print("Setting up data...")
afs, lbs = setup_data()
num_parts = 2
part_size = 500
parts = get_part_indexes(afs, num_parts, part_size)
test_size = 100
test_set = parts[1][:test_size]
train_set = parts[0]
print(f"Part size: {part_size} and {test_size=}")
print(f"Jaccard experiments")
distance = lambda x,y: jaccard(x, y, afs)
classifier=lambda x: lcl(x, lbs)
run_jaccard(train=train_set, test=test_set, distance=distance, classifier=classifier)
print("Euclid experiments")
distance = lambda x,y: adf(x, y, afs)
run_euclid(train=train_set, test=test_set, distance=distance, classifier=classifier)