Skip to main content

Library/framework for making predictions.

Project description

mydatapreprocessing

PyPI pyversions PyPI version Language grade: Python Build Status Documentation Status License: MIT codecov

Load data from web link or local file (json, csv, excel file, parquet, h5...), consolidate it and do preprocessing like resampling, standardization, string embedding, new columns derivation, feature extraction etc. based on configuration.

Library contain 3 modules.

Preprocessing

First - preprocessing load data, consolidate it and do the preprocessing. It contains functions like load_data, data_consolidation, preprocess_data, preprocess_data_inverse, add_frequency_columns, rolling_windows, add_derived_columns etc.

Example

import mydatapreprocessing.preprocessing as mdpp

data = "https://blockchain.info/unconfirmed-transactions?format=json"

# Load data from file or URL
data_loaded = mdpp.load_data(data, request_datatype_suffix=".json", predicted_table='txs')


#Some examples of other inputs to data_load function

# myarray_or_dataframe # Numpy array or Pandas.DataFrame
# r"/home/user/my.json" # Local file. The same with .parquet, .h5, .json or .xlsx. On windows it's necessary to use raw string - 'r' in front of string because of escape symbols \
# "https://yoururl/your.csv" # Web url (with suffix). Same with json.
# "https://blockchain.info/unconfirmed-transactions?format=json" # In this case you have to specify also 'request_datatype_suffix': "json", 'data_orientation': "index", 'predicted_table': 'txs',
# {'col_1': [3, 2, 1, 0], 'col_2': ['a', 'b', 'c', 'd']} # Dict with colums or rows (index) - necessary to setup data_orientation!


# You can use more files in list and data will be concatenated. It can be list of paths or list of python objects. Example:

# [{'col_1': 3, 'col_2': 'a'}, {'col_1': 0, 'col_2': 'd'}]  # List of records
# [np.random.randn(20, 3), np.random.randn(25, 3)]  # Dataframe same way
# ["https://raw.githubusercontent.com/jbrownlee/Datasets/master/daily-min-temperatures.csv", "https://raw.githubusercontent.com/jbrownlee/Datasets/master/daily-min-temperatures.csv"]  # List of URLs
# ["path/to/my1.csv", "path/to/my1.csv"]


# Transform various data into defined format - pandas dataframe - convert to numeric if possible, keep
# only numeric data and resample ifg configured. It return array, dataframe
data_consolidated = mdpp.data_consolidation(
    data_loaded, predicted_column="weight", data_orientation="index", remove_nans_threshold=0.9, remove_nans_or_replace='interpolate')

# You can add some extra informations to the data that can help (beware it can slow down the machine learning model)
to_be_extended = np.array([[0, 2] * 64, [0, 0, 0, 5] * 32]).T
extended = mdpp.add_frequency_columns(to_be_extended, window=8)


to_be_extended2 = pd.DataFrame([range(30), range(30, 60)]).T
extended2 = mdpp.add_derived_columns(to_be_extended2, differences=True, second_differences=True, multiplications=True,
                                    rolling_means=True, rolling_stds=True, mean_distances=True, window=10)

# Feature extraction is under development  :[

# Preprocess data. It return preprocessed data, but also last undifferenced value and scaler for inverse
# transformation, so unpack it with _
data_preprocessed, _, _ = mdpp.preprocess_data(data_consolidated, remove_outliers=True, smoothit=False,
                                              correlation_threshold=False, data_transform=False, standardizeit='standardize')

Inputs

Second module is inputs. It take tabular time series data and put it into format (input vector X, output vector y and input for predicted value x_input) that can be inserted into machine learning models for example on sklearn or tensorflow. It contain functions make_sequences, create_inputs and create_tests_outputs

Example for n_steps_in = 3 and n_steps_out = 1

From [[1], [2], [3], [4], [5], [6]]

Inputs: [[1, 2, 3], [2, 3, 4], [3, 4, 5]] Outputs [[4], [5], [6]]

Also multivariate data can be used.

import mydatapreprocessing as mdp

data = np.array([[1, 2, 3, 4, 5, 6, 7, 8], [9, 10, 11, 12 ,13, 14 ,15, 16], [17 ,18 ,19, 20, 21, 22, 23, 24]]).T
X, y, x_input, _ = mdp.inputs.make_sequences(data, n_steps_in= 2, n_steps_out=3)

# This example create from such a array:

# data = array([[1, 9, 17],
#               [2, 10, 18],
#               [3, 11, 19],
#               [4, 12, 20],
#               [5, 13, 21],
#               [6, 14, 22],
#               [7, 15, 23],
#               [8, 16, 24]])

# Such a results (data are serialized).

# X = array([[1, 2, 3, 9, 10, 11, 17, 18, 19],
#            [2, 3, 4, 10, 11, 12, 18, 19, 20],
#            [3, 4, 5, 11, 12, 13, 19, 20, 21],
#            [4, 5, 6, 12, 13, 14, 20, 21, 22]])

# y = array([[4, 5],
#            [5, 6],
#            [6, 7],
#            [7, 8]]

# x_input = array([[ 6,  7,  8, 14, 15, 16, 22, 23, 24]])

Third module is generatedata. It generate some basic data like sin, ramp random. In the future, it will also import some real datasets for models KPI.

Example

import mydatapreprocessing as mdp

data = mdp.generatedata.gen_sin(1000)

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for mydatapreprocessing, version 1.1.22
Filename, size File type Python version Upload date Hashes
Filename, size mydatapreprocessing-1.1.22-py3.7.egg (55.4 kB) File type Egg Python version 3.7 Upload date Hashes View
Filename, size mydatapreprocessing-1.1.22-py3-none-any.whl (27.9 kB) File type Wheel Python version py3 Upload date Hashes View
Filename, size mydatapreprocessing-1.1.22.tar.gz (25.1 kB) File type Source Python version None Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page