pyFTS.hyperparam package¶
Module contents¶
Submodules¶
pyFTS.hyperparam.Util module¶
Common facilities for hyperparameter optimization
- pyFTS.hyperparam.Util.create_hyperparam_tables(conn)[source]¶
Create a sqlite3 table designed to store benchmark results.
- Parameters
conn – a sqlite3 database connection
- pyFTS.hyperparam.Util.insert_hyperparam(data, conn)[source]¶
Insert benchmark data on database
- Parameters
data – a tuple with the benchmark data with format:
Dataset: Identify on which dataset the dataset was performed Tag: a user defined word that indentify a benchmark set Model: FTS model Transformation: The name of data transformation, if one was used mf: membership function Order: the order of the FTS method Partitioner: UoD partitioning scheme Partitions: Number of partitions alpha: alpha cut lags: lags Measure: accuracy measure Value: the measure value
- Parameters
conn – a sqlite3 database connection
- Returns
pyFTS.hyperparam.GridSearch module¶
pyFTS.hyperparam.Evolutionary module¶
Distributed Evolutionary Hyperparameter Optimization (DEHO) for MVFTS
- pyFTS.hyperparam.Evolutionary.GeneticAlgorithm(dataset, **kwargs)[source]¶
Genetic algoritm for Distributed Evolutionary Hyperparameter Optimization (DEHO)
- Parameters
dataset – The time series to optimize the FTS
ngen – An integer value with the maximum number of generations, default value: 30
mgen – An integer value with the maximum number of generations without improvement to stop, default value 7
npop – An integer value with the population size, default value: 20
pcross – A float value between 0 and 1 with the probability of crossover, default: .5
psel – A float value between 0 and 1 with the probability of selection, default: .5
pmut – A float value between 0 and 1 with the probability of mutation, default: .3
fts_method – The FTS method to optimize
parameters – dict with model specific arguments for fts_method
elitism – A boolean value indicating if the best individual must always survive to next population
initial_operator – a function that receives npop and return a random population with size npop
evalutation_operator – a function that receives a dataset and an individual and return its fitness
selection_operator – a function that receives the whole population and return a selected individual
crossover_operator – a function that receives the whole population and return a descendent individual
mutation_operator – a function that receives one individual and return a changed individual
window_size – An integer value with the the length of scrolling window for train/test on dataset
train_rate – A float value between 0 and 1 with the train/test split ([0,1])
increment_rate – A float value between 0 and 1 with the the increment of the scrolling window, relative to the window_size ([0,1])
collect_statistics – A boolean value indicating to collect statistics for each generation
distributed – A value indicating it the execution will be local and sequential (distributed=False), or parallel and distributed (distributed=’dispy’ or distributed=’spark’)
cluster – If distributed=’dispy’ the list of cluster nodes, else if distributed=’spark’ it is the master node
- Returns
the best genotype
- pyFTS.hyperparam.Evolutionary.crossover(population, **kwargs)[source]¶
Crossover operation between two parents
- Parameters
population – the original population
- Returns
a genotype
- pyFTS.hyperparam.Evolutionary.double_tournament(population, **kwargs)[source]¶
Double tournament selection strategy.
- Parameters
population –
- Returns
- pyFTS.hyperparam.Evolutionary.elitism(population, new_population, **kwargs)[source]¶
Elitism operation, always select the best individual of the population and discard the worst
- Parameters
population –
new_population –
- Returns
- pyFTS.hyperparam.Evolutionary.evaluate(dataset, individual, **kwargs)[source]¶
Evaluate an individual using a sliding window cross validation over the dataset.
- Parameters
dataset – Evaluation dataset
individual – genotype to be tested
window_size – The length of scrolling window for train/test on dataset
train_rate – The train/test split ([0,1])
increment_rate – The increment of the scrolling window, relative to the window_size ([0,1])
parameters – dict with model specific arguments for fit method.
- Returns
a tuple (len_lags, rmse) with the parsimony fitness value and the accuracy fitness value
- pyFTS.hyperparam.Evolutionary.execute(datasetname, dataset, **kwargs)[source]¶
Batch execution of Distributed Evolutionary Hyperparameter Optimization (DEHO) for monovariate methods
- Parameters
datasetname –
dataset – The time series to optimize the FTS
file –
experiments –
distributed –
ngen – An integer value with the maximum number of generations, default value: 30
mgen – An integer value with the maximum number of generations without improvement to stop, default value 7
npop – An integer value with the population size, default value: 20
pcross – A float value between 0 and 1 with the probability of crossover, default: .5
psel – A float value between 0 and 1 with the probability of selection, default: .5
pmut – A float value between 0 and 1 with the probability of mutation, default: .3
fts_method – The FTS method to optimize
parameters – dict with model specific arguments for fts_method
elitism – A boolean value indicating if the best individual must always survive to next population
initial_operator – a function that receives npop and return a random population with size npop
random_individual – create an random genotype
evalutation_operator – a function that receives a dataset and an individual and return its fitness
selection_operator – a function that receives the whole population and return a selected individual
crossover_operator – a function that receives the whole population and return a descendent individual
mutation_operator – a function that receives one individual and return a changed individual
window_size – An integer value with the the length of scrolling window for train/test on dataset
train_rate – A float value between 0 and 1 with the train/test split ([0,1])
increment_rate – A float value between 0 and 1 with the the increment of the scrolling window, relative to the window_size ([0,1])
collect_statistics – A boolean value indicating to collect statistics for each generation
distributed – A value indicating it the execution will be local and sequential (distributed=False), or parallel and distributed (distributed=’dispy’ or distributed=’spark’)
cluster – If distributed=’dispy’ the list of cluster nodes, else if distributed=’spark’ it is the master node
- Returns
the best genotype
- pyFTS.hyperparam.Evolutionary.genotype(mf, npart, partitioner, order, alpha, lags, f1, f2)[source]¶
Create the individual genotype
- Parameters
mf – membership function
npart – number of partitions
partitioner – partitioner method
order – model order
alpha – alpha-cut
lags – array with lag indexes
f1 – accuracy fitness value
f2 – parsimony fitness value
- Returns
the genotype, a dictionary with all hyperparameters
- pyFTS.hyperparam.Evolutionary.initial_population(n, **kwargs)[source]¶
Create a random population of size n
- Parameters
n – the size of the population
- Returns
a list with n random individuals
- pyFTS.hyperparam.Evolutionary.lag_crossover2(best, worst)[source]¶
Cross over two lag genes
- Parameters
best – best genotype
worst – worst genotype
- Returns
a tuple (order, lags)
- pyFTS.hyperparam.Evolutionary.mutation(individual, **kwargs)[source]¶
Mutation operator
- Parameters
individual – an individual genotype
pmut – individual probability o
- Returns
- pyFTS.hyperparam.Evolutionary.mutation_lags(lags, order)[source]¶
Mutation operation for lags gene
- Parameters
lags –
order –
- Returns
- pyFTS.hyperparam.Evolutionary.phenotype(individual, train, fts_method, parameters={}, **kwargs)[source]¶
Instantiate the genotype, creating a fitted model with the genotype hyperparameters
- Parameters
individual – a genotype
train – the training dataset
fts_method – the FTS method
parameters – dict with model specific arguments for fit method.
- Returns
a fitted FTS model
- pyFTS.hyperparam.Evolutionary.process_experiment(fts_method, result, datasetname, conn)[source]¶
Persist the results of an DEHO execution in sqlite database (best hyperparameters) and json file (generation statistics)
- Parameters
fts_method –
result –
datasetname –
conn –
- Returns