alu/tra-analysis
1
0
Fork 0
mirror of https://github.com/titanscouting/tra-analysis.git synced 2024-12-27 01:59:08 +00:00
Code Issues Packages Projects Releases Wiki Activity

analysis.py v 1.1.3.002

Browse Source
This commit is contained in:
art 2019-10-04 10:34:31 -05:00
parent 1cdeab4b6b
commit 3a17ac5154
1 changed files with 22 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

39
data analysis/analysis/analysis.py
View File

@ -7,10 +7,12 @@
# current benchmark of optimization: 1.33 times faster # current benchmark of optimization: 1.33 times faster
# setup: # setup:
__version__ = "1.1.3.001" __version__ = "1.1.3.002"
# changelog should be viewed using print(analysis.__changelog__) # changelog should be viewed using print(analysis.__changelog__)
__changelog__ = """changelog: __changelog__ = """changelog:
1.1.3.002:
- renamed regression class to Regression, regression_engine() to regression gliko2_engine class to Gliko2
1.1.3.001: 1.1.3.001:
- changed glicko2() to return tuple instead of array - changed glicko2() to return tuple instead of array
1.1.3.000: 1.1.3.000:
@ -161,11 +163,14 @@ __all__ = [
'z_score', 'z_score',
'z_normalize', 'z_normalize',
'histo_analysis', 'histo_analysis',
'regression_engine', 'regression',
'elo',
'gliko2',
'r_squared', 'r_squared',
'mse', 'mse',
'rms', 'rms',
'regression' 'Regression',
'Gliko2'
# all statistics functions left out due to integration in other functions # all statistics functions left out due to integration in other functions
] ]
@ -243,27 +248,27 @@ def histo_analysis(hist_data):
return basic_stats(derivative)[0], basic_stats(derivative)[3] return basic_stats(derivative)[0], basic_stats(derivative)[3]
@jit(forceobj=True) @jit(forceobj=True)
def regression_engine(device, inputs, outputs, args, loss = torch.nn.MSELoss(), _iterations = 10000, lr = 0.01): def regression(device, inputs, outputs, args, loss = torch.nn.MSELoss(), _iterations = 10000, lr = 0.01):
regressions = [] regressions = []
if 'cuda' in device: if 'cuda' in device:
regression.set_device(device) Regression.set_device(device)
if 'linear' in args: if 'linear' in args:
model = regression.SGDTrain(regression.LinearRegKernel(len(inputs)), torch.tensor(inputs).to(torch.float).cuda(), torch.tensor([outputs]).to(torch.float).cuda(), iterations=_iterations, learning_rate=lr, return_losses=True) model = Regression.SGDTrain(Regression.LinearRegKernel(len(inputs)), torch.tensor(inputs).to(torch.float).cuda(), torch.tensor([outputs]).to(torch.float).cuda(), iterations=_iterations, learning_rate=lr, return_losses=True)
regressions.append([model[0].parameters, model[1][::-1][0]]) regressions.append([model[0].parameters, model[1][::-1][0]])
if 'log' in args: if 'log' in args:
model = regression.SGDTrain(regression.LogRegKernel(len(inputs)), torch.tensor(inputs).to(torch.float).cuda(), torch.tensor(outputs).to(torch.float).cuda(), iterations=_iterations, learning_rate=lr, return_losses=True) model = Regression.SGDTrain(Regression.LogRegKernel(len(inputs)), torch.tensor(inputs).to(torch.float).cuda(), torch.tensor(outputs).to(torch.float).cuda(), iterations=_iterations, learning_rate=lr, return_losses=True)
regressions.append([model[0].parameters, model[1][::-1][0]]) regressions.append([model[0].parameters, model[1][::-1][0]])
if 'exp' in args: if 'exp' in args:
model = regression.SGDTrain(regression.ExpRegKernel(len(inputs)), torch.tensor(inputs).to(torch.float).cuda(), torch.tensor(outputs).to(torch.float).cuda(), iterations=_iterations, learning_rate=lr, return_losses=True) model = Regression.SGDTrain(Regression.ExpRegKernel(len(inputs)), torch.tensor(inputs).to(torch.float).cuda(), torch.tensor(outputs).to(torch.float).cuda(), iterations=_iterations, learning_rate=lr, return_losses=True)
regressions.append([model[0].parameters, model[1][::-1][0]]) regressions.append([model[0].parameters, model[1][::-1][0]])
#if 'poly' in args: #if 'poly' in args:
@ -272,26 +277,26 @@ def regression_engine(device, inputs, outputs, args, loss = torch.nn.MSELoss(),
if 'sig' in args: if 'sig' in args:
model = regression.SGDTrain(regression.SigmoidalRegKernelArthur(len(inputs)), torch.tensor(inputs).to(torch.float).cuda(), torch.tensor(outputs).to(torch.float).cuda(), iterations=_iterations, learning_rate=lr, return_losses=True) model = Regression.SGDTrain(Regression.SigmoidalRegKernelArthur(len(inputs)), torch.tensor(inputs).to(torch.float).cuda(), torch.tensor(outputs).to(torch.float).cuda(), iterations=_iterations, learning_rate=lr, return_losses=True)
regressions.append([model[0].parameters, model[1][::-1][0]]) regressions.append([model[0].parameters, model[1][::-1][0]])
else: else:
regression.set_device(device) Regression.set_device(device)
if 'linear' in args: if 'linear' in args:
model = regression.SGDTrain(regression.LinearRegKernel(len(inputs)), torch.tensor(inputs).to(torch.float), torch.tensor(outputs).to(torch.float), iterations=_iterations, learning_rate=lr, return_losses=True) model = Regression.SGDTrain(Regression.LinearRegKernel(len(inputs)), torch.tensor(inputs).to(torch.float), torch.tensor(outputs).to(torch.float), iterations=_iterations, learning_rate=lr, return_losses=True)
regressions.append([model[0].parameters, model[1][::-1][0]]) regressions.append([model[0].parameters, model[1][::-1][0]])
if 'log' in args: if 'log' in args:
model = regression.SGDTrain(regression.LogRegKernel(len(inputs)), torch.tensor(inputs).to(torch.float), torch.tensor(outputs).to(torch.float), iterations=_iterations, learning_rate=lr, return_losses=True) model = Regression.SGDTrain(Regression.LogRegKernel(len(inputs)), torch.tensor(inputs).to(torch.float), torch.tensor(outputs).to(torch.float), iterations=_iterations, learning_rate=lr, return_losses=True)
regressions.append([model[0].parameters, model[1][::-1][0]]) regressions.append([model[0].parameters, model[1][::-1][0]])
if 'exp' in args: if 'exp' in args:
model = regression.SGDTrain(regression.ExpRegKernel(len(inputs)), torch.tensor(inputs).to(torch.float), torch.tensor(outputs).to(torch.float), iterations=_iterations, learning_rate=lr, return_losses=True) model = Regression.SGDTrain(Regression.ExpRegKernel(len(inputs)), torch.tensor(inputs).to(torch.float), torch.tensor(outputs).to(torch.float), iterations=_iterations, learning_rate=lr, return_losses=True)
regressions.append([model[0].parameters, model[1][::-1][0]]) regressions.append([model[0].parameters, model[1][::-1][0]])
#if 'poly' in args: #if 'poly' in args:
@ -300,7 +305,7 @@ def regression_engine(device, inputs, outputs, args, loss = torch.nn.MSELoss(),
if 'sig' in args: if 'sig' in args:
model = regression.SGDTrain(regression.SigmoidalRegKernelArthur(len(inputs)), torch.tensor(inputs).to(torch.float), torch.tensor(outputs).to(torch.float), iterations=_iterations, learning_rate=lr, return_losses=True) model = Regression.SGDTrain(Regression.SigmoidalRegKernelArthur(len(inputs)), torch.tensor(inputs).to(torch.float), torch.tensor(outputs).to(torch.float), iterations=_iterations, learning_rate=lr, return_losses=True)
regressions.append([model[0].parameters, model[1][::-1][0]]) regressions.append([model[0].parameters, model[1][::-1][0]])
return regressions return regressions
@ -315,7 +320,7 @@ def elo(starting_score, opposing_scores, observed, N, K):
@jit(forceobj=True) @jit(forceobj=True)
def gliko2(starting_score, starting_rd, starting_vol, opposing_scores, opposing_rd, observations): def gliko2(starting_score, starting_rd, starting_vol, opposing_scores, opposing_rd, observations):
player = gliko2_engine(rating = starting_score, rd = starting_rd, vol = starting_vol) player = Gliko2(rating = starting_score, rd = starting_rd, vol = starting_vol)
player.update_player([x for x in opposing_scores], [x for x in opposing_rd], observations) player.update_player([x for x in opposing_scores], [x for x in opposing_rd], observations)
@ -356,7 +361,7 @@ def variance(data):
return np.var(data) return np.var(data)
class regression: class Regression:
# Titan Robotics Team 2022: CUDA-based Regressions Module # Titan Robotics Team 2022: CUDA-based Regressions Module
# Written by Arthur Lu & Jacob Levine # Written by Arthur Lu & Jacob Levine
@ -576,7 +581,7 @@ class regression:
optim.step() optim.step()
return kernel return kernel
class gliko2_engine: class Gliko2:
_tau = 0.5 _tau = 0.5