def main():
vcf_path, bed_path, expression_matrix_path, covariate_numerical, covariate_categorical, covariate_file_path, output_dir = argument_parser()
# extract basic information from input files
vcf_sample, vcf_var, expression_sample, expression_gene, cov_sample, cov_num, cov_used_num = file_info(vcf_path, expression_matrix_path, covariate_numerical, covariate_categorical, covariate_file_path)
print("VCF file:", vcf_sample, "samples and", vcf_var, "variants")
print("Expression matrix:", expression_sample, "samples and", expression_gene, "genes")
print("Covariate file:", cov_sample, "samples and", cov_num, "covariates;", cov_used_num, "covariates used")
# map samples to genes, write bin expression file for linear regression
annotation.main(vcf_path, bed_path, expression_matrix_path, output_dir)
# build linear regression model and perform t test, write output summary file
print('buiding regression model...')
regression.main(output_dir, covariate_numerical, covariate_categorical, covariate_file_path)
print('done')
return 0
def main(run_nums, data_dir, write_dir):
print('Setting up files for runs ' + str(run_nums))
for run_num in run_nums.split(' '):
print('Run: ' + str(run_num))
# Set up filtered data files
print('Creating filtered files...')
filters.main(True, True, run_num, data_dir, write_dir)
# Run RF regression on each run to obtain ground truth files
print('Creating ground truth files...')
regression.main(run_num, None, 'RF', data_dir, write_dir, 2)
print('\n')
print('Done')
def post(self):
next_values = {}
users = []
files = self.request.files
for file_name in files:
updateData(file_name, files[file_name][0]['body'].decode("utf-8"))
#execute regression
predict_value = regression.main('data/'+file_name)/MULTIPLICAND
next_values[file_name] = predict_value
users.append(file_name.split('_')[0])
calculateNextMove(next_values, users)
def post(self):
next_values = {}
users = []
files = self.request.files
for file_name in files:
updateData(file_name, files[file_name][0]['body'].decode("utf-8"))
#execute regression
predict_value = regression.main('data/' + file_name) / MULTIPLICAND
next_values[file_name] = predict_value
users.append(file_name.split('_')[0])
calculateNextMove(next_values, users)
def predict_regression():
collection = request.get_json()['collection']
features = request.get_json()['features']
target = request.get_json()['target']
inputType = request.get_json()['inputType']
ml_result = regression.main(collection, features, target, inputType)
return jsonify({
"Linear_Regression": ml_result['Linear_Regression'].tolist(),
"RandomForest": ml_result['RandomForest'].tolist(),
"GradientBoosting": ml_result['GradientBoosting'].tolist(),
"files": ml_result.index.tolist()
})
def main():
# Title
st.title("AlphaAI")
# Sidebar
activities = [
"Home", "Dataset Explorer", "ML Classifiers", "ML Regression",
"News Classification", "Text Summarizer",
"Real World Data Distribution", "Vision API"
]
choice = st.sidebar.selectbox("Choose Activity", activities)
if choice == "Home":
st.header(
'Empowering companies to jumpstart AI and generate real-world value'
)
st.subheader(
'Use exponential technologies to your advantage and lead your industry with confidence through innovation.'
)
image = Image.open('images/img0.jpg')
st.image(image, use_column_width=True, caption='Data Mining')
if choice == "Dataset Explorer":
st.subheader("Dataset Explorer")
dataset_analysis.main()
if choice == "Real World Data Distribution":
geo_climate.main()
if choice == "ML Regression":
regression.main()
if choice == "ML Classifiers":
classification.main()
if choice == "Vision API":
vision_api.main()
if choice == "Text Summarizer":
text_summ.main()
if choice == "News Classification":
newsclass.main()
# -*- coding: utf-8 -*-
#
# Copyright (c) 2017, the cclib development team
#
# This file is part of cclib (http://cclib.github.io) and is distributed under
# the terms of the BSD 3-Clause License.
"""This script runs the regression framework in the cclib-data repostiory."""
from __future__ import print_function
import os
import sys
if __name__ == "__main__":
# Assume the cclib-data repository is cloned in this directory.
regression_dir = os.path.join("..", "data", "regression")
sys.path.append(regression_dir)
import regression
opt_traceback = "--traceback" in sys.argv
opt_status = "--status" in sys.argv
# This can be used to limit the programs we want to run regressions for.
which = [arg for arg in sys.argv[1:] if not arg in ["--status", "--traceback"]]
regression.main(which, opt_traceback, opt_status, regression_dir)
# -*- coding: utf-8 -*-
"""
Created on Mon Feb 25 14:13:45 2019
@author: drape
"""
print('this should print')
import bank_boost as bb
import bagging as bag
import random_forest as rf
import regression as reg
print('does this print?')
bb.main()
bag.main()
rf.main()
reg.main()
def main(companyName, companySymbol, pred_date, weights, companyDict={}):
#initializing the neuron class
neural_network = NeuralNetwork()
#print("Beginning Randomly Generated Weights: ")
#print(neural_network.synaptic_weights)
if len(weights) == 0:
#training data consisting of 4 examples--3 input values and 1 output
training_examples = []
training_examples_outputs = []
companyDict = {
} # Avoid Slowdown from Yahoo and Nasdaq during training. Cache the results
for i in range(1, 13):
example = []
# set up dates
latestDate = datetime.datetime(2018, i, 2)
if i == 12:
endDate = datetime.datetime(2019, 1, 1)
else:
endDate = datetime.datetime(2018, i + 1, 1)
sentiment_analysis_tuple = SA.getSentiment(companySymbol,
companyName, latestDate,
endDate, companyDict)
example.append(sentiment_analysis_tuple[0])
# example.append(float(sentiment_analysis_tuple[1]/100))
companyDict = sentiment_analysis_tuple[2]
while True:
try:
regression_tuple = regression.main(companySymbol, endDate)
if regression_tuple[1] == None:
#print('No data for {}. Trying the next market day'.format(endDate.isoformat()))
endDate = endDate.replace(year=endDate.year,
month=endDate.month,
day=endDate.day + 1)
else:
break
except:
#print('Markets were closed on {}. Trying the next day this month'.format(endDate.isoformat()))
endDate = endDate.replace(year=endDate.year,
month=endDate.month,
day=endDate.day + 1)
#print(regression_tuple)
example.append(regression_tuple[0])
training_examples_outputs.append(regression_tuple[1]) #truth value
training_examples.append(example)
training_inputs = np.array(training_examples)
#print(training_inputs)
#training_inputs = np.array([[0,0,1],
# [1,1,1],
# [1,0,1],
# [0,1,1]])
#print(training_examples_outputs)
training_outputs = np.array([training_examples_outputs]).T
#training_outputs = np.array([[0,1,1,0]]).T
#training taking place
neural_network.train(training_inputs, training_outputs, 15000)
else:
neural_network.synaptic_weights = weights
#print("Ending Weights After Training: ")
#print(neural_network.synaptic_weights)
pred_back = pred_date.replace(year=pred_date.year,
month=pred_date.month,
day=pred_date.day - 2)
input_features = []
sentiment_analysis_tuple = SA.getSentiment(companySymbol, companyName,
pred_back, pred_date,
companyDict)
#print(sentiment_analysis_tuple)
input_features.append(sentiment_analysis_tuple[0])
#input_features.append(float(sentiment_analysis_tuple[1]/100))
#print(sentiment_analysis_tuple[1])
regression_tuple = regression.main(companySymbol, pred_date)
input_features.append(regression_tuple[0])
actual = regression_tuple[1]
#print(pred_date.isoformat())
#print("\t Sentiment", sentiment_analysis_tuple[0])
#if sentiment_analysis_tuple[0] > SENTIMENT_BOUNDARY:
# return (1, neural_network.synaptic_weights, companyDict)
#else:
# return (0, neural_network.synaptic_weights, companyDict)
print("\n Considering New Situation: " + str(input_features))
prediction = neural_network.think(np.array(input_features))
print("Trial for day ", pred_date.isoformat())
print("\t Prediction: ", prediction)
print("\t Actual Change: ", actual)
if prediction < actual + .1 and prediction > actual - .1:
return (1, neural_network.synaptic_weights, companyDict)
else:
return (0, neural_network.synaptic_weights, companyDict)
'''
Created on 20 mrt. 2013
@author: Erik Vandeputte
'''
import kmeans_assign
import regression
import time
RESULTSFILE = "results.txt"
clusters = [100, 200, 250, 300, 400, 450, 500, 700]
start_time = time.time()
f = open(RESULTSFILE, 'a')
for num_clusters in clusters:
print 'perform kmeans for %d clusters' % num_clusters
#kmeans_full.main(num_clusters)
#assign (hard and soft)
kmeans_assign.main(True, num_clusters)
kmeans_assign.main(False, num_clusters)
#perform regression(hard and soft)
mse_hard, alpha_hard = regression.main(True, num_clusters)
mse_soft, alpha_soft = regression.main(False, num_clusters)
f.write("%s\t%s\t%s\t%s\n" %
(str(num_clusters), 'hard', str(mse_hard), str(alpha_hard)))
f.write("%s\t%s\t%s\t%s\n" %
(str(num_clusters), 'soft', str(mse_soft), str(alpha_soft)))
f.flush()
f.write("running this script took %.2f seconds" % (time.time() - start_time))
f.close()
def main():
import argparse
parser = argparse.ArgumentParser(description='Neural Network framework.')
parser.add_argument(
'action',
choices=['regression', 'classification'],
help='Choose mode either \'regression\' or \'classification\'.')
parser.add_argument(
'activation',
choices=['sigmoid', 'relu', 'tanh'],
help='Choose mode either \'sigmoid\' or \'relu\' or \'tanh\'.')
parser.add_argument('--train_filename',
type=str,
help='Name of a file containing training data',
required=False)
parser.add_argument('--test_filename',
type=str,
help='Name of a file containing testing data')
parser.add_argument(
'--create_nn',
nargs='*',
type=int,
help=
'When creating a nn from scratch; number of neurons for each layer',
required=False)
parser.add_argument('--save_nn',
type=str,
help='Name of a file to save trained model to.')
parser.add_argument('--savefig_filename',
type=str,
help='Name of a file to save plot to.')
parser.add_argument('-e',
'--number_of_epochs',
type=int,
help='Number of epochs (iterations) for the NN to run',
required=False,
default=10000)
parser.add_argument('--read_nn',
type=str,
help='When reading existing nn from a file; filename')
parser.add_argument(
'-v',
'--visualize_every',
type=int,
help='How ofter (every n iterations) print neuron\'s weights.',
required=False)
parser.add_argument('--l_rate',
type=float,
help='Learning rate',
required=False,
default=0.001)
parser.add_argument('--seed',
type=int,
help='Random seed int',
required=False,
default=1)
parser.add_argument('--biases', dest='biases', action='store_true')
parser.add_argument('--no_biases', dest='biases', action='store_false')
parser.set_defaults(biases=True)
args = parser.parse_args()
# Seed the random number generator
random.seed(args.seed)
if args.create_nn is None and args.read_nn is None:
print('Either \'--create_nn\' or \'--read_nn\' has to be provided.')
exit(1)
if args.train_filename is None and args.save_nn is not None:
print(
'\'--save_nn\' cannot be provided when \'--train_filename\' is not provided.'
)
exit(1)
if args.train_filename is None and args.create_nn is not None:
print(
'\'--create_nn\' cannot be provided when \'--train_filename\' is not provided.'
)
exit(1)
if args.activation == 'sigmoid':
from util import sigmoid, sigmoid_derivative
activation_f, activation_f_derivative = sigmoid, sigmoid_derivative
elif args.activation == 'relu':
from util import reLu, reLu_derivative
activation_f, activation_f_derivative = reLu, reLu_derivative
elif args.activation == 'tanh':
from util import tanh, tanh_derivative
activation_f, activation_f_derivative = tanh, tanh_derivative
else:
print(
'Sorry, second positional argument has to be either \'sigmoid\' or \'relu\' or \'tanh\'.'
)
exit(1)
if args.action == 'regression':
import regression
regression.main(args.train_filename, args.test_filename,
args.create_nn, args.save_nn, args.read_nn,
args.number_of_epochs, args.visualize_every,
args.l_rate, args.savefig_filename, activation_f,
activation_f_derivative)
elif args.action == 'classification':
import classification
classification.main(args.train_filename, args.test_filename,
args.create_nn, args.save_nn, args.read_nn,
args.number_of_epochs, args.visualize_every,
args.l_rate, args.biases, activation_f,
activation_f_derivative)
else:
print(
'Sorry, first positional argument has to be either \'regression\' or \'classification\'.'
)
exit(1)
'''
Created on 20 mrt. 2013
@author: Erik Vandeputte
'''
import kmeans_assign
import regression
import time
RESULTSFILE = "results.txt"
clusters = [100,200,250,300,400,450,500,700]
start_time = time.time()
f = open(RESULTSFILE,'a')
for num_clusters in clusters:
print 'perform kmeans for %d clusters' %num_clusters
#kmeans_full.main(num_clusters)
#assign (hard and soft)
kmeans_assign.main(True,num_clusters)
kmeans_assign.main(False,num_clusters)
#perform regression(hard and soft)
mse_hard,alpha_hard = regression.main(True, num_clusters)
mse_soft,alpha_soft = regression.main(False, num_clusters)
f.write("%s\t%s\t%s\t%s\n" % (str(num_clusters),'hard',str(mse_hard),str(alpha_hard)))
f.write("%s\t%s\t%s\t%s\n" % (str(num_clusters),'soft',str(mse_soft),str(alpha_soft)))
f.flush()
f.write("running this script took %.2f seconds" % (time.time() - start_time))
f.close()