def __init__(self):
self.sim = simulation.Simulation()
self.caches = []
for _ in range(16):
new_cache = mem.cached(algorithm=mem.LFU,
max_size=len(c.ACTION_LIST)**2,
custom_key_maker=self._custom_keys)(
self.getNextState)
self.caches.append(new_cache)
def cached(user_function=None,
max_size=500000,
ttl=None,
algorithm=CachingAlgorithmFlag.LRU,
thread_safe=True,
order_independent=False):
"""
Entry point for the configuration of the memoization cache (max_size=500000)
"""
return memoization.cached(user_function, max_size, ttl, algorithm,
thread_safe, order_independent)
def _cache_in_memory(self):
@self._update_signature(signature(self.func).parameters.values())
def make_custom_key(*args, **kwargs):
key = self._get_data_key(**kwargs)
return key
f = cached(ttl=self.expired_time,
custom_key_maker=make_custom_key)(self.func)
@wraps(f)
def wrapper(*args, **kwargs):
return f(*args, **kwargs)
return wrapper
def cached_wrappers(user_function=None,
max_size=0,
ttl=None,
algorithm=CachingAlgorithmFlag.LRU,
thread_safe=True,
order_independent=False):
"""
Entry point for the configuration of the memoization cache
To find the optimal max-size value:
from sharadar.util.cache import wrappers
for wrapper in wrappers:
print(wrapper, wrapper.cache_info())
"""
wrapper = memoization.cached(user_function, max_size, ttl, algorithm,
thread_safe, order_independent)
wrappers.append(wrapper)
return wrapper
def fit_mice_execute(X: DataFrame, iterations: int = 1) -> Any:
kernel = MultipleImputedKernel(X, save_all_iterations=True, datasets=1)
kernel.mice(iterations, verbose=True)
return kernel
fit_mice_execute_cached = memory.cache(fit_mice_execute)
def fit_mice_memory_cached_(X: DataFrame, iterations: int = 1) -> Any:
print('MICE cache miss')
return fit_mice_execute(X, iterations)
fit_mice_execute_memory_cached = cached(fit_mice_memory_cached_, custom_key_maker=fit_mice_hash)
class MiceForest(BaseEstimator, TransformerMixin):
def __init__(self, iterations: int = 1):
self.kernel = None
self.iterations = iterations
def fit(self, X, y=None, **fit_params):
self.kernel = fit_mice_execute_memory_cached(X)
return self
def transform(self, X, y=None):
print('MICE transforming')
data = self.kernel.impute_new_data(X).complete_data(self.iterations)