resistics.config module¶
Module containing the resistics configuration
The configuration is an essential part of a resistics environment. It defines many dependencies, such as which data readers to use for time series data or calibration data and processing options.
Configuration allows users to insert their own dependencies and processors to work with data.
Configurations can be saved to and loaded from JSON files.
- pydantic model resistics.config.Configuration[source]¶
Bases:
resistics.common.ResisticsModelThe resistics configuration
Configuration can be customised by users who wish to use their own custom processes for certain steps. In most cases, customisation will be for:
Implementing new time data readers
Implementing readers for specific calibration formats
Implementing time data processors
Implementing spectra data processors
Adding new features to extract from the data
Examples
Frequently, configuration will be used to change data readers.
>>> from resistics.letsgo import get_default_configuration >>> config = get_default_configuration() >>> config.name 'default' >>> for tr in config.time_readers: ... tr.summary() { 'name': 'TimeReaderAscii', 'apply_scalings': True, 'extension': '.txt', 'delimiter': None, 'n_header': 0 } { 'name': 'TimeReaderNumpy', 'apply_scalings': True, 'extension': '.npy' } >>> config.sensor_calibrator.summary() { 'name': 'SensorCalibrator', 'chans': None, 'readers': [ { 'name': 'SensorCalibrationJSON', 'extension': '.json', 'file_str': 'IC_$sensor$extension' } ] }
To change these, it’s best to make a new configuration with a different name
>>> from resistics.letsgo import Configuration >>> from resistics.time import TimeReaderNumpy >>> config = Configuration(name="myconfig", time_readers=[TimeReaderNumpy(apply_scalings=False)]) >>> for tr in config.time_readers: ... tr.summary() { 'name': 'TimeReaderNumpy', 'apply_scalings': False, 'extension': '.npy' }
Or for the sensor calibration
>>> from resistics.calibrate import SensorCalibrator, SensorCalibrationTXT >>> calibration_reader = SensorCalibrationTXT(file_str="lemi120_IC_$serial$extension") >>> calibrator = SensorCalibrator(chans=["Hx", "Hy", "Hz"], readers=[calibration_reader]) >>> config = Configuration(name="myconfig", sensor_calibrator=calibrator) >>> config.sensor_calibrator.summary() { 'name': 'SensorCalibrator', 'chans': ['Hx', 'Hy', 'Hz'], 'readers': [ { 'name': 'SensorCalibrationTXT', 'extension': '.TXT', 'file_str': 'lemi120_IC_$serial$extension' } ] }
As a final example, create a configuration which used targetted windowing instead of specified window sizes
>>> from resistics.letsgo import Configuration >>> from resistics.window import WindowerTarget >>> config = Configuration(name="window_target", windower=WindowerTarget(target=500)) >>> config.name 'window_target' >>> config.windower.summary() { 'name': 'WindowerTarget', 'target': 500, 'min_size': 64, 'olap_proportion': 0.25 }
Show JSON schema
{ "title": "Configuration", "description": "The resistics configuration\n\nConfiguration can be customised by users who wish to use their own custom\nprocesses for certain steps. In most cases, customisation will be for:\n\n- Implementing new time data readers\n- Implementing readers for specific calibration formats\n- Implementing time data processors\n- Implementing spectra data processors\n- Adding new features to extract from the data\n\nExamples\n--------\nFrequently, configuration will be used to change data readers.\n\n>>> from resistics.letsgo import get_default_configuration\n>>> config = get_default_configuration()\n>>> config.name\n'default'\n>>> for tr in config.time_readers:\n... tr.summary()\n{\n 'name': 'TimeReaderAscii',\n 'apply_scalings': True,\n 'extension': '.txt',\n 'delimiter': None,\n 'n_header': 0\n}\n{\n 'name': 'TimeReaderNumpy',\n 'apply_scalings': True,\n 'extension': '.npy'\n}\n>>> config.sensor_calibrator.summary()\n{\n 'name': 'SensorCalibrator',\n 'chans': None,\n 'readers': [\n {\n 'name': 'SensorCalibrationJSON',\n 'extension': '.json',\n 'file_str': 'IC_$sensor$extension'\n }\n ]\n}\n\nTo change these, it's best to make a new configuration with a different name\n\n>>> from resistics.letsgo import Configuration\n>>> from resistics.time import TimeReaderNumpy\n>>> config = Configuration(name=\"myconfig\", time_readers=[TimeReaderNumpy(apply_scalings=False)])\n>>> for tr in config.time_readers:\n... tr.summary()\n{\n 'name': 'TimeReaderNumpy',\n 'apply_scalings': False,\n 'extension': '.npy'\n}\n\nOr for the sensor calibration\n\n>>> from resistics.calibrate import SensorCalibrator, SensorCalibrationTXT\n>>> calibration_reader = SensorCalibrationTXT(file_str=\"lemi120_IC_$serial$extension\")\n>>> calibrator = SensorCalibrator(chans=[\"Hx\", \"Hy\", \"Hz\"], readers=[calibration_reader])\n>>> config = Configuration(name=\"myconfig\", sensor_calibrator=calibrator)\n>>> config.sensor_calibrator.summary()\n{\n 'name': 'SensorCalibrator',\n 'chans': ['Hx', 'Hy', 'Hz'],\n 'readers': [\n {\n 'name': 'SensorCalibrationTXT',\n 'extension': '.TXT',\n 'file_str': 'lemi120_IC_$serial$extension'\n }\n ]\n}\n\nAs a final example, create a configuration which used targetted windowing\ninstead of specified window sizes\n\n>>> from resistics.letsgo import Configuration\n>>> from resistics.window import WindowerTarget\n>>> config = Configuration(name=\"window_target\", windower=WindowerTarget(target=500))\n>>> config.name\n'window_target'\n>>> config.windower.summary()\n{\n 'name': 'WindowerTarget',\n 'target': 500,\n 'min_size': 64,\n 'olap_proportion': 0.25\n}", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "time_readers": { "title": "Time Readers", "default": [ { "name": "TimeReaderAscii", "apply_scalings": true, "extension": ".txt", "delimiter": null, "n_header": 0 }, { "name": "TimeReaderNumpy", "apply_scalings": true, "extension": ".npy" } ], "type": "array", "items": { "$ref": "#/definitions/TimeReader" } }, "time_processors": { "title": "Time Processors", "default": [ { "name": "InterpolateNans" }, { "name": "RemoveMean" } ], "type": "array", "items": { "$ref": "#/definitions/TimeProcess" } }, "dec_setup": { "title": "Dec Setup", "default": { "name": "DecimationSetup", "n_levels": 8, "per_level": 5, "min_samples": 256, "div_factor": 2, "eval_freqs": null }, "allOf": [ { "$ref": "#/definitions/DecimationSetup" } ] }, "decimator": { "title": "Decimator", "default": { "name": "Decimator", "resample": true, "max_single_factor": 3 }, "allOf": [ { "$ref": "#/definitions/Decimator" } ] }, "win_setup": { "title": "Win Setup", "default": { "name": "WindowSetup", "min_size": 128, "min_olap": 32, "win_factor": 4, "olap_proportion": 0.25, "min_n_wins": 5, "win_sizes": null, "olap_sizes": null }, "allOf": [ { "$ref": "#/definitions/WindowSetup" } ] }, "windower": { "title": "Windower", "default": { "name": "Windower" }, "allOf": [ { "$ref": "#/definitions/Windower" } ] }, "fourier": { "title": "Fourier", "default": { "name": "FourierTransform", "win_fnc": [ "kaiser", 14 ], "detrend": "linear", "workers": -2 }, "allOf": [ { "$ref": "#/definitions/FourierTransform" } ] }, "spectra_processors": { "title": "Spectra Processors", "default": [], "type": "array", "items": { "$ref": "#/definitions/SpectraProcess" } }, "evals": { "title": "Evals", "default": { "name": "EvaluationFreqs" }, "allOf": [ { "$ref": "#/definitions/EvaluationFreqs" } ] }, "sensor_calibrator": { "title": "Sensor Calibrator", "default": { "name": "SensorCalibrator", "chans": null, "readers": [ { "name": "SensorCalibrationJSON", "extension": ".json", "file_str": "IC_$sensor$extension" } ] }, "allOf": [ { "$ref": "#/definitions/Calibrator" } ] }, "tf": { "title": "Tf", "default": { "name": "ImpedanceTensor", "variation": "default", "out_chans": [ "Ex", "Ey" ], "in_chans": [ "Hx", "Hy" ], "cross_chans": [ "Hx", "Hy" ], "n_out": 2, "n_in": 2, "n_cross": 2 }, "allOf": [ { "$ref": "#/definitions/TransferFunction" } ] }, "regression_preparer": { "title": "Regression Preparer", "default": { "name": "RegressionPreparerGathered" }, "allOf": [ { "$ref": "#/definitions/RegressionPreparerGathered" } ] }, "solver": { "title": "Solver", "default": { "name": "SolverScikitTheilSen", "fit_intercept": false, "normalize": false, "n_jobs": -2, "max_subpopulation": 2000, "n_subsamples": null }, "allOf": [ { "$ref": "#/definitions/Solver" } ] } }, "required": [ "name" ], "definitions": { "TimeReader": { "title": "TimeReader", "description": "Base class for resistics processes\n\nResistics processes perform operations on data (including read and write\noperations). Each time a ResisticsProcess child class is run, it should add\na process record to the dataset", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "apply_scalings": { "title": "Apply Scalings", "default": true, "type": "boolean" }, "extension": { "title": "Extension", "type": "string" } } }, "TimeProcess": { "title": "TimeProcess", "description": "Parent class for processing time data", "type": "object", "properties": { "name": { "title": "Name", "type": "string" } } }, "DecimationSetup": { "title": "DecimationSetup", "description": "Process to calculate decimation parameters\n\nParameters\n----------\nn_levels : int, optional\n Number of decimation levels, by default 8\nper_level : int, optional\n Number of frequencies per level, by default 5\nmin_samples : int, optional\n Number of samples under which to quit decimating\ndiv_factor : int, optional\n Minimum division factor for decimation, by default 2.\neval_freqs : Optional[List[float]], optional\n Explicit definition of evaluation frequencies as a flat list, by\n default None. Must be of size n_levels * per_level\n\nExamples\n--------\n>>> from resistics.decimate import DecimationSetup\n>>> dec_setup = DecimationSetup(n_levels=3, per_level=2)\n>>> dec_params = dec_setup.run(128)\n>>> print(dec_params.to_dataframe())\n 0 1 fs factors increments\ndecimation level\n0 32.0 22.627417 128.0 1 1\n1 16.0 11.313708 64.0 2 2\n2 8.0 5.656854 32.0 4 2", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "n_levels": { "title": "N Levels", "default": 8, "type": "integer" }, "per_level": { "title": "Per Level", "default": 5, "type": "integer" }, "min_samples": { "title": "Min Samples", "default": 256, "type": "integer" }, "div_factor": { "title": "Div Factor", "default": 2, "type": "integer" }, "eval_freqs": { "title": "Eval Freqs", "type": "array", "items": { "type": "number" } } } }, "Decimator": { "title": "Decimator", "description": "Decimate the time data into multiple levels\n\nThere are two options for decimation, using time data Resample or using\ntime data Decimate. The default is to use Resample.", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "resample": { "title": "Resample", "default": true, "type": "boolean" }, "max_single_factor": { "title": "Max Single Factor", "default": 3, "minimum": 3, "type": "integer" } } }, "WindowSetup": { "title": "WindowSetup", "description": "Setup WindowParameters\n\nWindowSetup outputs the WindowParameters to use for windowing decimated\ntime data.\n\nWindow parameters are simply the window and overlap sizes for each\ndecimation level.\n\nParameters\n----------\nmin_size : int, optional\n Minimum window size, by default 128\nmin_olap : int, optional\n Minimum overlap size, by default 32\nwin_factor : int, optional\n Window factor, by default 4. Window sizes are calculated by sampling\n frequency / 4 to ensure sufficient frequency resolution. If the\n sampling frequency is small, window size will be adjusted to\n min_size\nolap_proportion : float, optional\n The proportion of the window size to use as the overlap, by default\n 0.25. For example, for a window size of 128, the overlap would be\n 0.25 * 128 = 32\nmin_n_wins : int, optional\n The minimum number of windows needed in a decimation level, by\n default 5\nwin_sizes : Optional[List[int]], optional\n Explicit define window sizes, by default None. Must have the same\n length as number of decimation levels\nolap_sizes : Optional[List[int]], optional\n Explicitly define overlap sizes, by default None. Must have the same\n length as number of decimation levels\n\nExamples\n--------\nGenerate decimation and windowing parameters for data sampled at 0.05 Hz or\n20 seconds sampling period\n\n>>> from resistics.decimate import DecimationSetup\n>>> from resistics.window import WindowSetup\n>>> dec_params = DecimationSetup(n_levels=3, per_level=3).run(0.05)\n>>> dec_params.summary()\n{\n 'fs': 0.05,\n 'n_levels': 3,\n 'per_level': 3,\n 'min_samples': 256,\n 'eval_freqs': [\n 0.0125,\n 0.008838834764831844,\n 0.00625,\n 0.004419417382415922,\n 0.003125,\n 0.002209708691207961,\n 0.0015625,\n 0.0011048543456039805,\n 0.00078125\n ],\n 'dec_factors': [1, 2, 8],\n 'dec_increments': [1, 2, 4],\n 'dec_fs': [0.05, 0.025, 0.00625]\n}\n>>> win_params = WindowSetup().run(dec_params.n_levels, dec_params.dec_fs)\n>>> win_params.summary()\n{\n 'n_levels': 3,\n 'min_n_wins': 5,\n 'win_sizes': [128, 128, 128],\n 'olap_sizes': [32, 32, 32]\n}\n\nWindow parameters can also be explicitly defined\n\n>>> from resistics.decimate import DecimationSetup\n>>> from resistics.window import WindowSetup\n>>> dec_setup = DecimationSetup(n_levels=3, per_level=3)\n>>> dec_params = dec_setup.run(0.05)\n>>> win_setup = WindowSetup(win_sizes=[1000, 578, 104])\n>>> win_params = win_setup.run(dec_params.n_levels, dec_params.dec_fs)\n>>> win_params.summary()\n{\n 'n_levels': 3,\n 'min_n_wins': 5,\n 'win_sizes': [1000, 578, 104],\n 'olap_sizes': [250, 144, 32]\n}", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "min_size": { "title": "Min Size", "default": 128, "type": "integer" }, "min_olap": { "title": "Min Olap", "default": 32, "type": "integer" }, "win_factor": { "title": "Win Factor", "default": 4, "type": "integer" }, "olap_proportion": { "title": "Olap Proportion", "default": 0.25, "type": "number" }, "min_n_wins": { "title": "Min N Wins", "default": 5, "type": "integer" }, "win_sizes": { "title": "Win Sizes", "type": "array", "items": { "type": "integer" } }, "olap_sizes": { "title": "Olap Sizes", "type": "array", "items": { "type": "integer" } } } }, "Windower": { "title": "Windower", "description": "Windows DecimatedData\n\nThis is the primary window making process for resistics and should be used\nwhen alignment of windows with a site or across sites is required.\n\nThis method uses numpy striding to produce window views into the decimated\ndata.\n\nSee Also\n--------\nWindowerTarget : A windower to make a target number of windows\n\nExamples\n--------\nThe Windower windows a DecimatedData object given a reference time and some\nwindow parameters.\n\nThere's quite a few imports needed for this example. Begin by doing the\nimports, defining a reference time and generating random decimated data.\n\n>>> from resistics.sampling import to_datetime\n>>> from resistics.testing import decimated_data_linear\n>>> from resistics.window import WindowSetup, Windower\n>>> dec_data = decimated_data_linear(fs=128)\n>>> ref_time = dec_data.metadata.first_time\n>>> print(dec_data.to_string())\n<class 'resistics.decimate.DecimatedData'>\n fs dt n_samples first_time last_time\nlevel\n0 2048.0 0.000488 16384 2021-01-01 00:00:00 2021-01-01 00:00:07.99951171875\n1 512.0 0.001953 4096 2021-01-01 00:00:00 2021-01-01 00:00:07.998046875\n2 128.0 0.007812 1024 2021-01-01 00:00:00 2021-01-01 00:00:07.9921875\n\nNext, initialise the window parameters. For this example, use small windows,\nwhich will make inspecting them easier.\n\n>>> win_params = WindowSetup(win_sizes=[16,16,16], min_olap=4).run(dec_data.metadata.n_levels, dec_data.metadata.fs)\n>>> win_params.summary()\n{\n 'n_levels': 3,\n 'min_n_wins': 5,\n 'win_sizes': [16, 16, 16],\n 'olap_sizes': [4, 4, 4]\n}\n\nPerform the windowing. This actually creates views into the decimated data\nusing the numpy.lib.stride_tricks.sliding_window_view function. The shape\nfor a data array at a decimation level is: n_wins x n_chans x win_size. The\ninformation about each level is also in the levels_metadata attribute of\nWindowedMetadata.\n\n>>> win_data = Windower().run(ref_time, win_params, dec_data)\n>>> win_data.data[0].shape\n(1365, 2, 16)\n>>> for level_metadata in win_data.metadata.levels_metadata:\n... level_metadata.summary()\n{\n 'fs': 2048.0,\n 'n_wins': 1365,\n 'win_size': 16,\n 'olap_size': 4,\n 'index_offset': 0\n}\n{\n 'fs': 512.0,\n 'n_wins': 341,\n 'win_size': 16,\n 'olap_size': 4,\n 'index_offset': 0\n}\n{\n 'fs': 128.0,\n 'n_wins': 85,\n 'win_size': 16,\n 'olap_size': 4,\n 'index_offset': 0\n}\n\nLet's look at an example of data from the first decimation level for the\nfirst channel. This is simply a linear set of data ranging from 0...16_383.\n\n>>> dec_data.data[0][0]\narray([ 0, 1, 2, ..., 16381, 16382, 16383])\n\nInspecting the first few windows shows they are as expected including the\noverlap.\n\n>>> win_data.data[0][0, 0]\narray([ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15])\n>>> win_data.data[0][1, 0]\narray([12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27])\n>>> win_data.data[0][2, 0]\narray([24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39])", "type": "object", "properties": { "name": { "title": "Name", "type": "string" } } }, "FourierTransform": { "title": "FourierTransform", "description": "Perform a Fourier transform of the windowed data\n\nThe processor is inspired by the scipy.signal.stft function which performs\na similar process and involves a Fourier transform along the last axis of\nthe windowed data.\n\nParameters\n----------\nwin_fnc : Union[str, Tuple[str, float]]\n The window to use before performing the FFT, by default (\"kaiser\", 14)\ndetrend : Union[str, None]\n Type of detrending to apply before performing FFT, by default linear\n detrend. Setting to None will not apply any detrending to the data prior\n to the FFT\nworkers : int\n The number of CPUs to use, by default max - 2\n\nExamples\n--------\nThis example will get periodic decimated data, perfrom windowing and run the\nFourier transform on the windowed data.\n\n.. plot::\n :width: 90%\n\n >>> import matplotlib.pyplot as plt\n >>> import numpy as np\n >>> from resistics.testing import decimated_data_periodic\n >>> from resistics.window import WindowSetup, Windower\n >>> from resistics.spectra import FourierTransform\n >>> frequencies = {\"chan1\": [870, 590, 110, 32, 12], \"chan2\": [480, 375, 210, 60, 45]}\n >>> dec_data = decimated_data_periodic(frequencies, fs=128)\n >>> dec_data.metadata.chans\n ['chan1', 'chan2']\n >>> print(dec_data.to_string())\n <class 'resistics.decimate.DecimatedData'>\n fs dt n_samples first_time last_time\n level\n 0 2048.0 0.000488 16384 2021-01-01 00:00:00 2021-01-01 00:00:07.99951171875\n 1 512.0 0.001953 4096 2021-01-01 00:00:00 2021-01-01 00:00:07.998046875\n 2 128.0 0.007812 1024 2021-01-01 00:00:00 2021-01-01 00:00:07.9921875\n\n Perform the windowing\n\n >>> win_params = WindowSetup().run(dec_data.metadata.n_levels, dec_data.metadata.fs)\n >>> win_data = Windower().run(dec_data.metadata.first_time, win_params, dec_data)\n\n And then the Fourier transform. By default, the data will be (linearly)\n detrended and mutliplied by a Kaiser window prior to the Fourier\n transform\n\n >>> spec_data = FourierTransform().run(win_data)\n\n For plotting of magnitude, let's stack the spectra\n\n >>> freqs_0 = spec_data.metadata.levels_metadata[0].freqs\n >>> data_0 = np.absolute(spec_data.data[0]).mean(axis=0)\n >>> freqs_1 = spec_data.metadata.levels_metadata[1].freqs\n >>> data_1 = np.absolute(spec_data.data[1]).mean(axis=0)\n >>> freqs_2 = spec_data.metadata.levels_metadata[2].freqs\n >>> data_2 = np.absolute(spec_data.data[2]).mean(axis=0)\n\n Now plot\n\n >>> plt.subplot(3,1,1) # doctest: +SKIP\n >>> plt.plot(freqs_0, data_0[0], label=\"chan1\") # doctest: +SKIP\n >>> plt.plot(freqs_0, data_0[1], label=\"chan2\") # doctest: +SKIP\n >>> plt.grid()\n >>> plt.title(\"Decimation level 0\") # doctest: +SKIP\n >>> plt.legend() # doctest: +SKIP\n >>> plt.subplot(3,1,2) # doctest: +SKIP\n >>> plt.plot(freqs_1, data_1[0], label=\"chan1\") # doctest: +SKIP\n >>> plt.plot(freqs_1, data_1[1], label=\"chan2\") # doctest: +SKIP\n >>> plt.grid()\n >>> plt.title(\"Decimation level 1\") # doctest: +SKIP\n >>> plt.legend() # doctest: +SKIP\n >>> plt.subplot(3,1,3) # doctest: +SKIP\n >>> plt.plot(freqs_2, data_2[0], label=\"chan1\") # doctest: +SKIP\n >>> plt.plot(freqs_2, data_2[1], label=\"chan2\") # doctest: +SKIP\n >>> plt.grid()\n >>> plt.title(\"Decimation level 2\") # doctest: +SKIP\n >>> plt.legend() # doctest: +SKIP\n >>> plt.xlabel(\"Frequency\") # doctest: +SKIP\n >>> plt.tight_layout() # doctest: +SKIP\n >>> plt.show() # doctest: +SKIP", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "win_fnc": { "title": "Win Fnc", "default": [ "kaiser", 14 ], "anyOf": [ { "type": "string" }, { "type": "array", "items": [ { "type": "string" }, { "type": "number" } ] } ] }, "detrend": { "title": "Detrend", "default": "linear", "type": "string" }, "workers": { "title": "Workers", "default": -2, "type": "integer" } } }, "SpectraProcess": { "title": "SpectraProcess", "description": "Parent class for spectra processes", "type": "object", "properties": { "name": { "title": "Name", "type": "string" } } }, "EvaluationFreqs": { "title": "EvaluationFreqs", "description": "Calculate the spectra values at the evaluation frequencies\n\nThis is done using linear interpolation in the complex domain\n\nExample\n-------\nThe example will show interpolation to evaluation frequencies on a very\nsimple example. Begin by generating some example spectra data.\n\n>>> from resistics.decimate import DecimationSetup\n>>> from resistics.spectra import EvaluationFreqs\n>>> from resistics.testing import spectra_data_basic\n>>> spec_data = spectra_data_basic()\n>>> spec_data.metadata.n_levels\n1\n>>> spec_data.metadata.chans\n['chan1']\n>>> spec_data.metadata.levels_metadata[0].summary()\n{\n 'fs': 180.0,\n 'n_wins': 2,\n 'win_size': 20,\n 'olap_size': 5,\n 'index_offset': 0,\n 'n_freqs': 10,\n 'freqs': [0.0, 10.0, 20.0, 30.0, 40.0, 50.0, 60.0, 70.0, 80.0, 90.0]\n}\n\nThe spectra data has only a single channel and a single level which has 2\nwindows. Now define our evaluation frequencies.\n\n>>> eval_freqs = [1, 12, 23, 34, 45, 56, 67, 78, 89]\n>>> dec_setup = DecimationSetup(n_levels=1, per_level=9, eval_freqs=eval_freqs)\n>>> dec_params = dec_setup.run(spec_data.metadata.fs[0])\n>>> dec_params.summary()\n{\n 'fs': 180.0,\n 'n_levels': 1,\n 'per_level': 9,\n 'min_samples': 256,\n 'eval_freqs': [1.0, 12.0, 23.0, 34.0, 45.0, 56.0, 67.0, 78.0, 89.0],\n 'dec_factors': [1],\n 'dec_increments': [1],\n 'dec_fs': [180.0]\n}\n\nNow calculate the spectra at the evaluation frequencies\n\n>>> eval_data = EvaluationFreqs().run(dec_params, spec_data)\n>>> eval_data.metadata.levels_metadata[0].summary()\n{\n 'fs': 180.0,\n 'n_wins': 2,\n 'win_size': 20,\n 'olap_size': 5,\n 'index_offset': 0,\n 'n_freqs': 9,\n 'freqs': [1.0, 12.0, 23.0, 34.0, 45.0, 56.0, 67.0, 78.0, 89.0]\n}\n\nTo double check everything is as expected, let's compare the data. Comparing\nwindow 1 gives\n\n>>> print(spec_data.data[0][0, 0])\n[0.+0.j 1.+1.j 2.+2.j 3.+3.j 4.+4.j 5.+5.j 6.+6.j 7.+7.j 8.+8.j 9.+9.j]\n>>> print(eval_data.data[0][0, 0])\n[0.1+0.1j 1.2+1.2j 2.3+2.3j 3.4+3.4j 4.5+4.5j 5.6+5.6j 6.7+6.7j 7.8+7.8j\n 8.9+8.9j]\n\nAnd window 2\n\n>>> print(spec_data.data[0][1, 0])\n[-1. +1.j 0. +2.j 1. +3.j 2. +4.j 3. +5.j 4. +6.j 5. +7.j 6. +8.j\n 7. +9.j 8.+10.j]\n>>> print(eval_data.data[0][1, 0])\n[-0.9+1.1j 0.2+2.2j 1.3+3.3j 2.4+4.4j 3.5+5.5j 4.6+6.6j 5.7+7.7j\n 6.8+8.8j 7.9+9.9j]", "type": "object", "properties": { "name": { "title": "Name", "type": "string" } } }, "Calibrator": { "title": "Calibrator", "description": "Parent class for a calibrator", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "chans": { "title": "Chans", "type": "array", "items": { "type": "string" } } } }, "TransferFunction": { "title": "TransferFunction", "description": "Define a generic transfer function\n\nThis class is a describes generic transfer function, including:\n\n- The output channels for the transfer function\n- The input channels for the transfer function\n- The cross channels for the transfer function\n\nThe cross channels are the channels that will be used to calculate out the\ncross powers for the regression.\n\nThis generic parent class has no implemented plotting function. However,\nchild classes may have a plotting function as different transfer functions\nmay need different types of plots.\n\n.. note::\n\n Users interested in writing a custom transfer function should inherit\n from this generic Transfer function\n\nSee Also\n--------\nImpandanceTensor : Transfer function for the MT impedance tensor\nTipper : Transfer function for the MT tipper\n\nExamples\n--------\nA generic example\n\n>>> tf = TransferFunction(variation=\"example\", out_chans=[\"bye\", \"see you\", \"ciao\"], in_chans=[\"hello\", \"hi_there\"])\n>>> print(tf.to_string())\n| bye | | bye_hello bye_hi_there | | hello |\n| see you | = | see you_hello see you_hi_there | | hi_there |\n| ciao | | ciao_hello ciao_hi_there |\n\nCombining the impedance tensor and the tipper into one TransferFunction\n\n>>> tf = TransferFunction(variation=\"combined\", out_chans=[\"Ex\", \"Ey\"], in_chans=[\"Hx\", \"Hy\", \"Hz\"])\n>>> print(tf.to_string())\n| Ex | | Ex_Hx Ex_Hy Ex_Hz | | Hx |\n| Ey | = | Ey_Hx Ey_Hy Ey_Hz | | Hy |\n | Hz |", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "variation": { "title": "Variation", "default": "generic", "maxLength": 16, "type": "string" }, "out_chans": { "title": "Out Chans", "type": "array", "items": { "type": "string" } }, "in_chans": { "title": "In Chans", "type": "array", "items": { "type": "string" } }, "cross_chans": { "title": "Cross Chans", "type": "array", "items": { "type": "string" } }, "n_out": { "title": "N Out", "type": "integer" }, "n_in": { "title": "N In", "type": "integer" }, "n_cross": { "title": "N Cross", "type": "integer" } }, "required": [ "out_chans", "in_chans" ] }, "RegressionPreparerGathered": { "title": "RegressionPreparerGathered", "description": "Regression preparer for gathered data\n\nIn nearly all cases, this is the regresson preparer to use. As input, it\nrequires GatheredData.", "type": "object", "properties": { "name": { "title": "Name", "type": "string" } } }, "Solver": { "title": "Solver", "description": "General resistics solver", "type": "object", "properties": { "name": { "title": "Name", "type": "string" } } } } }
- field name: str [Required]¶
The name of the configuration
- field time_readers: List[resistics.time.TimeReader] = [TimeReaderAscii(name='TimeReaderAscii', apply_scalings=True, extension='.txt', delimiter=None, n_header=0), TimeReaderNumpy(name='TimeReaderNumpy', apply_scalings=True, extension='.npy')]¶
Time readers in the configuration
- field time_processors: List[resistics.time.TimeProcess] = [InterpolateNans(name='InterpolateNans'), RemoveMean(name='RemoveMean')]¶
List of time processors to run
- field dec_setup: resistics.decimate.DecimationSetup = DecimationSetup(name='DecimationSetup', n_levels=8, per_level=5, min_samples=256, div_factor=2, eval_freqs=None)¶
Process to calculate decimation parameters
- field decimator: resistics.decimate.Decimator = Decimator(name='Decimator', resample=True, max_single_factor=3)¶
Process to decimate time data
- field win_setup: resistics.window.WindowSetup = WindowSetup(name='WindowSetup', min_size=128, min_olap=32, win_factor=4, olap_proportion=0.25, min_n_wins=5, win_sizes=None, olap_sizes=None)¶
Process to calculate windowing parameters
- field windower: resistics.window.Windower = Windower(name='Windower')¶
Process to window the decimated data
- field fourier: resistics.spectra.FourierTransform = FourierTransform(name='FourierTransform', win_fnc=('kaiser', 14), detrend='linear', workers=-2)¶
Process to perform the fourier transform
- field spectra_processors: List[resistics.spectra.SpectraProcess] = []¶
List of processors to run on spectra data
- field evals: resistics.spectra.EvaluationFreqs = EvaluationFreqs(name='EvaluationFreqs')¶
Process to get the spectra data at the evaluation frequencies
- field sensor_calibrator: resistics.calibrate.Calibrator = SensorCalibrator(name='SensorCalibrator', chans=None, readers=[SensorCalibrationJSON(name='SensorCalibrationJSON', extension='.json', file_str='IC_$sensor$extension')])¶
The sensor calibrator and associated calibration file readers
- field tf: resistics.transfunc.TransferFunction = ImpedanceTensor(name='ImpedanceTensor', variation='default', out_chans=['Ex', 'Ey'], in_chans=['Hx', 'Hy'], cross_chans=['Hx', 'Hy'], n_out=2, n_in=2, n_cross=2)¶
The transfer function to solve
- field regression_preparer: resistics.regression.RegressionPreparerGathered = RegressionPreparerGathered(name='RegressionPreparerGathered')¶
Process to prepare linear equations
- field solver: resistics.regression.Solver = SolverScikitTheilSen(name='SolverScikitTheilSen', fit_intercept=False, normalize=False, n_jobs=-2, max_subpopulation=2000, n_subsamples=None)¶
The solver to use to estimate the regression parameters
- resistics.config.get_default_configuration() resistics.config.Configuration[source]¶
Get the default configuration