resistics.regression module¶
The regression module provides functions and classes for the following:
Preparing gathered data for regression
Performing the linear regression
Resistics has built in solvers that use scikit learn models, namely
Ordinary least squares
RANSAC
TheilSen
These will perform well in many scenarios. However, the functionality available in resistics makes it possible to use custom solvers if required.
- pydantic model resistics.regression.RegressionInputMetadata[source]¶
Bases:
resistics.common.MetadataMetadata for regression input data, mainly to track processing history
Show JSON schema
{ "title": "RegressionInputMetadata", "description": "Metadata for regression input data, mainly to track processing history", "type": "object", "properties": { "contributors": { "title": "Contributors", "type": "object", "additionalProperties": { "anyOf": [ { "$ref": "#/definitions/SiteCombinedMetadata" }, { "$ref": "#/definitions/SpectraMetadata" } ] } }, "history": { "title": "History", "default": { "records": [] }, "allOf": [ { "$ref": "#/definitions/History" } ] } }, "required": [ "contributors" ], "definitions": { "ResisticsFile": { "title": "ResisticsFile", "description": "Required information for writing out a resistics file", "type": "object", "properties": { "created_on_local": { "title": "Created On Local", "type": "string", "format": "date-time" }, "created_on_utc": { "title": "Created On Utc", "type": "string", "format": "date-time" }, "version": { "title": "Version", "type": "string" } } }, "Record": { "title": "Record", "description": "Class to hold a record\n\nA record holds information about a process that was run. It is intended to\ntrack processes applied to data, allowing a process history to be saved\nalong with any datasets.\n\nExamples\n--------\nA simple example of creating a process record\n\n>>> from resistics.common import Record\n>>> messages = [\"message 1\", \"message 2\"]\n>>> record = Record(\n... creator={\"name\": \"example\", \"parameter1\": 15},\n... messages=messages,\n... record_type=\"example\"\n... )\n>>> record.summary()\n{\n 'time_local': '...',\n 'time_utc': '...',\n 'creator': {'name': 'example', 'parameter1': 15},\n 'messages': ['message 1', 'message 2'],\n 'record_type': 'example'\n}", "type": "object", "properties": { "time_local": { "title": "Time Local", "type": "string", "format": "date-time" }, "time_utc": { "title": "Time Utc", "type": "string", "format": "date-time" }, "creator": { "title": "Creator", "type": "object" }, "messages": { "title": "Messages", "type": "array", "items": { "type": "string" } }, "record_type": { "title": "Record Type", "type": "string" } }, "required": [ "creator", "messages", "record_type" ] }, "History": { "title": "History", "description": "Class for storing processing history\n\nParameters\n----------\nrecords : List[Record], optional\n List of records, by default []\n\nExamples\n--------\n>>> from resistics.testing import record_example1, record_example2\n>>> from resistics.common import History\n>>> record1 = record_example1()\n>>> record2 = record_example2()\n>>> history = History(records=[record1, record2])\n>>> history.summary()\n{\n 'records': [\n {\n 'time_local': '...',\n 'time_utc': '...',\n 'creator': {\n 'name': 'example1',\n 'a': 5,\n 'b': -7.0\n },\n 'messages': ['Message 1', 'Message 2'],\n 'record_type': 'process'\n },\n {\n 'time_local': '...',\n 'time_utc': '...',\n 'creator': {\n 'name': 'example2',\n 'a': 'parzen',\n 'b': -21\n },\n 'messages': ['Message 5', 'Message 6'],\n 'record_type': 'process'\n }\n ]\n}", "type": "object", "properties": { "records": { "title": "Records", "default": [], "type": "array", "items": { "$ref": "#/definitions/Record" } } } }, "SiteCombinedMetadata": { "title": "SiteCombinedMetadata", "description": "Metadata for combined data\n\nCombined metadata stores metadata for measurements that are combined from\na single site.", "type": "object", "properties": { "file_info": { "$ref": "#/definitions/ResisticsFile" }, "site_name": { "title": "Site Name", "type": "string" }, "fs": { "title": "Fs", "type": "number" }, "system": { "title": "System", "default": "", "type": "string" }, "serial": { "title": "Serial", "default": "", "type": "string" }, "wgs84_latitude": { "title": "Wgs84 Latitude", "default": -999.0, "type": "number" }, "wgs84_longitude": { "title": "Wgs84 Longitude", "default": -999.0, "type": "number" }, "easting": { "title": "Easting", "default": -999.0, "type": "number" }, "northing": { "title": "Northing", "default": -999.0, "type": "number" }, "elevation": { "title": "Elevation", "default": -999.0, "type": "number" }, "measurements": { "title": "Measurements", "type": "array", "items": { "type": "string" } }, "chans": { "title": "Chans", "type": "array", "items": { "type": "string" } }, "n_evals": { "title": "N Evals", "type": "integer" }, "eval_freqs": { "title": "Eval Freqs", "type": "array", "items": { "type": "number" } }, "histories": { "title": "Histories", "type": "object", "additionalProperties": { "$ref": "#/definitions/History" } } }, "required": [ "site_name", "fs", "chans", "n_evals", "eval_freqs", "histories" ] }, "ChanMetadata": { "title": "ChanMetadata", "description": "Channel metadata", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "data_files": { "title": "Data Files", "type": "array", "items": { "type": "string" } }, "chan_type": { "title": "Chan Type", "type": "string" }, "chan_source": { "title": "Chan Source", "type": "string" }, "sensor": { "title": "Sensor", "default": "", "type": "string" }, "serial": { "title": "Serial", "default": "", "type": "string" }, "gain1": { "title": "Gain1", "default": 1, "type": "number" }, "gain2": { "title": "Gain2", "default": 1, "type": "number" }, "scaling": { "title": "Scaling", "default": 1, "type": "number" }, "chopper": { "title": "Chopper", "default": false, "type": "boolean" }, "dipole_dist": { "title": "Dipole Dist", "default": 1, "type": "number" }, "sensor_calibration_file": { "title": "Sensor Calibration File", "default": "", "type": "string" }, "instrument_calibration_file": { "title": "Instrument Calibration File", "default": "", "type": "string" } }, "required": [ "name" ] }, "SpectraLevelMetadata": { "title": "SpectraLevelMetadata", "description": "Metadata for spectra of a windowed decimation level", "type": "object", "properties": { "fs": { "title": "Fs", "type": "number" }, "n_wins": { "title": "N Wins", "type": "integer" }, "win_size": { "title": "Win Size", "exclusiveMinimum": 0, "type": "integer" }, "olap_size": { "title": "Olap Size", "exclusiveMinimum": 0, "type": "integer" }, "index_offset": { "title": "Index Offset", "type": "integer" }, "n_freqs": { "title": "N Freqs", "type": "integer" }, "freqs": { "title": "Freqs", "type": "array", "items": { "type": "number" } } }, "required": [ "fs", "n_wins", "win_size", "olap_size", "index_offset", "n_freqs", "freqs" ] }, "SpectraMetadata": { "title": "SpectraMetadata", "description": "Metadata for spectra data", "type": "object", "properties": { "file_info": { "$ref": "#/definitions/ResisticsFile" }, "fs": { "title": "Fs", "type": "array", "items": { "type": "number" } }, "chans": { "title": "Chans", "type": "array", "items": { "type": "string" } }, "n_chans": { "title": "N Chans", "type": "integer" }, "n_levels": { "title": "N Levels", "type": "integer" }, "first_time": { "title": "First Time", "pattern": "%Y-%m-%d %H:%M:%S.%f_%o_%q_%v", "examples": [ "2021-01-01 00:00:00.000061_035156_250000_000000" ] }, "last_time": { "title": "Last Time", "pattern": "%Y-%m-%d %H:%M:%S.%f_%o_%q_%v", "examples": [ "2021-01-01 00:00:00.000061_035156_250000_000000" ] }, "system": { "title": "System", "default": "", "type": "string" }, "serial": { "title": "Serial", "default": "", "type": "string" }, "wgs84_latitude": { "title": "Wgs84 Latitude", "default": -999.0, "type": "number" }, "wgs84_longitude": { "title": "Wgs84 Longitude", "default": -999.0, "type": "number" }, "easting": { "title": "Easting", "default": -999.0, "type": "number" }, "northing": { "title": "Northing", "default": -999.0, "type": "number" }, "elevation": { "title": "Elevation", "default": -999.0, "type": "number" }, "chans_metadata": { "title": "Chans Metadata", "type": "object", "additionalProperties": { "$ref": "#/definitions/ChanMetadata" } }, "levels_metadata": { "title": "Levels Metadata", "type": "array", "items": { "$ref": "#/definitions/SpectraLevelMetadata" } }, "ref_time": { "title": "Ref Time", "pattern": "%Y-%m-%d %H:%M:%S.%f_%o_%q_%v", "examples": [ "2021-01-01 00:00:00.000061_035156_250000_000000" ] }, "history": { "title": "History", "default": { "records": [] }, "allOf": [ { "$ref": "#/definitions/History" } ] } }, "required": [ "fs", "chans", "n_levels", "first_time", "last_time", "chans_metadata", "levels_metadata", "ref_time" ] } } }
- field contributors: Dict[str, Union[resistics.gather.SiteCombinedMetadata, resistics.spectra.SpectraMetadata]] [Required]¶
Details about the data contributing to the regression input data
- field history: resistics.common.History = History(records=[])¶
The processing history
- class resistics.regression.RegressionInputData(metadata: resistics.regression.RegressionInputMetadata, tf: resistics.transfunc.TransferFunction, freqs: List[float], obs: List[Dict[str, numpy.ndarray]], preds: List[numpy.ndarray])[source]¶
Bases:
resistics.common.ResisticsDataClass to hold data that will be input into a solver
The purpose of regression input data is to provision for many different solvers and user written solvers.
The regression input data has the following key attributes:
freqs
obs
preds
The freqs attribute is a 1-D array of evaluation frequencies.
The obs attribute is a dictionary of dictionaries. The parent dictionary has a key of the evaluation frequency index. The secondary dictionary has key of output channel. The values in the secondary dictionary are the observations for that output channel and have 1-D size:
(n_wins x n_cross_chans x 2).
The factor of 2 is because the real and complex parts of each equation are separated into two equations to allow use of solvers that work exclusively on real data.
The preds attribute is a single level dictionary with key of evaluation frequency index and value of the predictors for the evaluation frequency. The predictors have 2-D shape:
(n_wins x n_cross_chans x 2) x (n_input_channels x 2).
The number of windows is multiplied by 2 for the same reason as the observations. The doubling of the input channels is because one is the predictor for the real part of that transfer function component and one is the predictor for the complex part of the transfer function component.
Considering the impedance tensor as an example with:
output channels Ex, Ey
input channels Hx, Hy
cross channels Hx, Hy
The below shows the arrays for the 0 index evaluation frequency:
Observations
Ex: [w1_crossHx_RE, w1_crossHx_IM, w1_crossHy_RE, w1_crossHy_IM]
Ey: [w1_crossHx_RE, w1_crossHx_IM, w1_crossHy_RE, w1_crossHy_IM]
Predictors Ex
w1_crossHx_RE: Zxx_RE Zxx_IM Zxy_RE Zxy_IM
w1_crossHx_IM: Zxx_RE Zxx_IM Zxy_RE Zxy_IM
w1_crossHy_RE: Zxx_RE Zxx_IM Zxy_RE Zxy_IM
w1_crossHy_IM: Zxx_RE Zxx_IM Zxy_RE Zxy_IM
Predictors Ey
w1_crossHx_RE: Zyx_RE Zyx_IM Zyy_RE Zyy_IM
w1_crossHx_IM: Zyx_RE Zyx_IM Zyy_RE Zyy_IM
w1_crossHy_RE: Zyx_RE Zyx_IM Zyy_RE Zyy_IM
w1_crossHy_IM: Zyx_RE Zyx_IM Zyy_RE Zyy_IM
Note that the predictors are the same regardless of the output channel, only the observations change.
- property n_freqs: int¶
Get the number of frequencies
- pydantic model resistics.regression.RegressionPreparerSpectra[source]¶
Bases:
resistics.common.ResisticsProcessPrepare regression data directly from spectra data
This can be useful for running a single measurement
Show JSON schema
{ "title": "RegressionPreparerSpectra", "description": "Prepare regression data directly from spectra data\n\nThis can be useful for running a single measurement", "type": "object", "properties": { "name": { "title": "Name", "type": "string" } } }
- run(tf: resistics.transfunc.TransferFunction, spec_data: resistics.spectra.SpectraData) resistics.regression.RegressionInputData[source]¶
Construct the linear equation for solving
- field name: Optional[str] [Required]¶
- Validated by
validate_name
- pydantic model resistics.regression.RegressionPreparerGathered[source]¶
Bases:
resistics.common.ResisticsProcessRegression preparer for gathered data
In nearly all cases, this is the regresson preparer to use. As input, it requires GatheredData.
Show JSON schema
{ "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" } } }
- run(tf: resistics.transfunc.TransferFunction, gathered_data: resistics.gather.GatheredData) resistics.regression.RegressionInputData[source]¶
Create the RegressionInputData
- Parameters
tf (TransferFunction) – The transfer function
gathered_data (GatheredData) – The gathered data
- Returns
Data that can be used as input into a solver
- Return type
- field name: Optional[str] [Required]¶
- Validated by
validate_name
- pydantic model resistics.regression.Solution[source]¶
Bases:
resistics.common.WriteableMetadataClass to hold a transfer function solution
Examples
>>> from resistics.testing import solution_mt >>> solution = solution_mt() >>> print(solution.tf.to_string()) | Ex | = | Ex_Hx Ex_Hy | | Hx | | Ey | | Ey_Hx Ey_Hy | | Hy | >>> solution.n_freqs 5 >>> solution.freqs [10.0, 20.0, 30.0, 40.0, 50.0] >>> solution.periods.tolist() [0.1, 0.05, 0.03333333333333333, 0.025, 0.02] >>> solution.components["ExHx"] Component(real=[1.0, 1.0, 2.0, 2.0, 3.0], imag=[5.0, 5.0, 4.0, 4.0, 3.0]) >>> solution.components["ExHy"] Component(real=[1.0, 2.0, 3.0, 4.0, 5.0], imag=[-5.0, -4.0, -3.0, -2.0, -1.0])
To get the components as an array, either get_component or subscripting be used
>>> solution["ExHy"] array([1.-5.j, 2.-4.j, 3.-3.j, 4.-2.j, 5.-1.j]) >>> solution["ab"] Traceback (most recent call last): ... ValueError: Component ab not found in ['ExHx', 'ExHy', 'EyHx', 'EyHy']
It is also possible to get the tensor values at a particular evaluation frequency
>>> solution.get_tensor(2) array([[ 2.+4.j, 3.-3.j], [-3.+3.j, -2.-4.j]])
Show JSON schema
{ "title": "Solution", "description": "Class to hold a transfer function solution\n\nExamples\n--------\n>>> from resistics.testing import solution_mt\n>>> solution = solution_mt()\n>>> print(solution.tf.to_string())\n| Ex | = | Ex_Hx Ex_Hy | | Hx |\n| Ey | | Ey_Hx Ey_Hy | | Hy |\n>>> solution.n_freqs\n5\n>>> solution.freqs\n[10.0, 20.0, 30.0, 40.0, 50.0]\n>>> solution.periods.tolist()\n[0.1, 0.05, 0.03333333333333333, 0.025, 0.02]\n>>> solution.components[\"ExHx\"]\nComponent(real=[1.0, 1.0, 2.0, 2.0, 3.0], imag=[5.0, 5.0, 4.0, 4.0, 3.0])\n>>> solution.components[\"ExHy\"]\nComponent(real=[1.0, 2.0, 3.0, 4.0, 5.0], imag=[-5.0, -4.0, -3.0, -2.0, -1.0])\n\nTo get the components as an array, either get_component or subscripting\nbe used\n\n>>> solution[\"ExHy\"]\narray([1.-5.j, 2.-4.j, 3.-3.j, 4.-2.j, 5.-1.j])\n>>> solution[\"ab\"]\nTraceback (most recent call last):\n...\nValueError: Component ab not found in ['ExHx', 'ExHy', 'EyHx', 'EyHy']\n\nIt is also possible to get the tensor values at a particular evaluation\nfrequency\n\n>>> solution.get_tensor(2)\narray([[ 2.+4.j, 3.-3.j],\n [-3.+3.j, -2.-4.j]])", "type": "object", "properties": { "file_info": { "$ref": "#/definitions/ResisticsFile" }, "tf": { "$ref": "#/definitions/TransferFunction" }, "freqs": { "title": "Freqs", "type": "array", "items": { "type": "number" } }, "components": { "title": "Components", "type": "object", "additionalProperties": { "$ref": "#/definitions/Component" } }, "history": { "$ref": "#/definitions/History" }, "contributors": { "title": "Contributors", "type": "object", "additionalProperties": { "anyOf": [ { "$ref": "#/definitions/SiteCombinedMetadata" }, { "$ref": "#/definitions/SpectraMetadata" } ] } } }, "required": [ "tf", "freqs", "components", "history", "contributors" ], "definitions": { "ResisticsFile": { "title": "ResisticsFile", "description": "Required information for writing out a resistics file", "type": "object", "properties": { "created_on_local": { "title": "Created On Local", "type": "string", "format": "date-time" }, "created_on_utc": { "title": "Created On Utc", "type": "string", "format": "date-time" }, "version": { "title": "Version", "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" ] }, "Component": { "title": "Component", "description": "Data class for a single component in a Transfer function\n\nExample\n-------\n>>> from resistics.transfunc import Component\n>>> component = Component(real=[1, 2, 3, 4, 5], imag=[-5, -4, -3, -2 , -1])\n>>> component.get_value(0)\n(1-5j)\n>>> component.to_numpy()\narray([1.-5.j, 2.-4.j, 3.-3.j, 4.-2.j, 5.-1.j])", "type": "object", "properties": { "real": { "title": "Real", "type": "array", "items": { "type": "number" } }, "imag": { "title": "Imag", "type": "array", "items": { "type": "number" } } }, "required": [ "real", "imag" ] }, "Record": { "title": "Record", "description": "Class to hold a record\n\nA record holds information about a process that was run. It is intended to\ntrack processes applied to data, allowing a process history to be saved\nalong with any datasets.\n\nExamples\n--------\nA simple example of creating a process record\n\n>>> from resistics.common import Record\n>>> messages = [\"message 1\", \"message 2\"]\n>>> record = Record(\n... creator={\"name\": \"example\", \"parameter1\": 15},\n... messages=messages,\n... record_type=\"example\"\n... )\n>>> record.summary()\n{\n 'time_local': '...',\n 'time_utc': '...',\n 'creator': {'name': 'example', 'parameter1': 15},\n 'messages': ['message 1', 'message 2'],\n 'record_type': 'example'\n}", "type": "object", "properties": { "time_local": { "title": "Time Local", "type": "string", "format": "date-time" }, "time_utc": { "title": "Time Utc", "type": "string", "format": "date-time" }, "creator": { "title": "Creator", "type": "object" }, "messages": { "title": "Messages", "type": "array", "items": { "type": "string" } }, "record_type": { "title": "Record Type", "type": "string" } }, "required": [ "creator", "messages", "record_type" ] }, "History": { "title": "History", "description": "Class for storing processing history\n\nParameters\n----------\nrecords : List[Record], optional\n List of records, by default []\n\nExamples\n--------\n>>> from resistics.testing import record_example1, record_example2\n>>> from resistics.common import History\n>>> record1 = record_example1()\n>>> record2 = record_example2()\n>>> history = History(records=[record1, record2])\n>>> history.summary()\n{\n 'records': [\n {\n 'time_local': '...',\n 'time_utc': '...',\n 'creator': {\n 'name': 'example1',\n 'a': 5,\n 'b': -7.0\n },\n 'messages': ['Message 1', 'Message 2'],\n 'record_type': 'process'\n },\n {\n 'time_local': '...',\n 'time_utc': '...',\n 'creator': {\n 'name': 'example2',\n 'a': 'parzen',\n 'b': -21\n },\n 'messages': ['Message 5', 'Message 6'],\n 'record_type': 'process'\n }\n ]\n}", "type": "object", "properties": { "records": { "title": "Records", "default": [], "type": "array", "items": { "$ref": "#/definitions/Record" } } } }, "SiteCombinedMetadata": { "title": "SiteCombinedMetadata", "description": "Metadata for combined data\n\nCombined metadata stores metadata for measurements that are combined from\na single site.", "type": "object", "properties": { "file_info": { "$ref": "#/definitions/ResisticsFile" }, "site_name": { "title": "Site Name", "type": "string" }, "fs": { "title": "Fs", "type": "number" }, "system": { "title": "System", "default": "", "type": "string" }, "serial": { "title": "Serial", "default": "", "type": "string" }, "wgs84_latitude": { "title": "Wgs84 Latitude", "default": -999.0, "type": "number" }, "wgs84_longitude": { "title": "Wgs84 Longitude", "default": -999.0, "type": "number" }, "easting": { "title": "Easting", "default": -999.0, "type": "number" }, "northing": { "title": "Northing", "default": -999.0, "type": "number" }, "elevation": { "title": "Elevation", "default": -999.0, "type": "number" }, "measurements": { "title": "Measurements", "type": "array", "items": { "type": "string" } }, "chans": { "title": "Chans", "type": "array", "items": { "type": "string" } }, "n_evals": { "title": "N Evals", "type": "integer" }, "eval_freqs": { "title": "Eval Freqs", "type": "array", "items": { "type": "number" } }, "histories": { "title": "Histories", "type": "object", "additionalProperties": { "$ref": "#/definitions/History" } } }, "required": [ "site_name", "fs", "chans", "n_evals", "eval_freqs", "histories" ] }, "ChanMetadata": { "title": "ChanMetadata", "description": "Channel metadata", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "data_files": { "title": "Data Files", "type": "array", "items": { "type": "string" } }, "chan_type": { "title": "Chan Type", "type": "string" }, "chan_source": { "title": "Chan Source", "type": "string" }, "sensor": { "title": "Sensor", "default": "", "type": "string" }, "serial": { "title": "Serial", "default": "", "type": "string" }, "gain1": { "title": "Gain1", "default": 1, "type": "number" }, "gain2": { "title": "Gain2", "default": 1, "type": "number" }, "scaling": { "title": "Scaling", "default": 1, "type": "number" }, "chopper": { "title": "Chopper", "default": false, "type": "boolean" }, "dipole_dist": { "title": "Dipole Dist", "default": 1, "type": "number" }, "sensor_calibration_file": { "title": "Sensor Calibration File", "default": "", "type": "string" }, "instrument_calibration_file": { "title": "Instrument Calibration File", "default": "", "type": "string" } }, "required": [ "name" ] }, "SpectraLevelMetadata": { "title": "SpectraLevelMetadata", "description": "Metadata for spectra of a windowed decimation level", "type": "object", "properties": { "fs": { "title": "Fs", "type": "number" }, "n_wins": { "title": "N Wins", "type": "integer" }, "win_size": { "title": "Win Size", "exclusiveMinimum": 0, "type": "integer" }, "olap_size": { "title": "Olap Size", "exclusiveMinimum": 0, "type": "integer" }, "index_offset": { "title": "Index Offset", "type": "integer" }, "n_freqs": { "title": "N Freqs", "type": "integer" }, "freqs": { "title": "Freqs", "type": "array", "items": { "type": "number" } } }, "required": [ "fs", "n_wins", "win_size", "olap_size", "index_offset", "n_freqs", "freqs" ] }, "SpectraMetadata": { "title": "SpectraMetadata", "description": "Metadata for spectra data", "type": "object", "properties": { "file_info": { "$ref": "#/definitions/ResisticsFile" }, "fs": { "title": "Fs", "type": "array", "items": { "type": "number" } }, "chans": { "title": "Chans", "type": "array", "items": { "type": "string" } }, "n_chans": { "title": "N Chans", "type": "integer" }, "n_levels": { "title": "N Levels", "type": "integer" }, "first_time": { "title": "First Time", "pattern": "%Y-%m-%d %H:%M:%S.%f_%o_%q_%v", "examples": [ "2021-01-01 00:00:00.000061_035156_250000_000000" ] }, "last_time": { "title": "Last Time", "pattern": "%Y-%m-%d %H:%M:%S.%f_%o_%q_%v", "examples": [ "2021-01-01 00:00:00.000061_035156_250000_000000" ] }, "system": { "title": "System", "default": "", "type": "string" }, "serial": { "title": "Serial", "default": "", "type": "string" }, "wgs84_latitude": { "title": "Wgs84 Latitude", "default": -999.0, "type": "number" }, "wgs84_longitude": { "title": "Wgs84 Longitude", "default": -999.0, "type": "number" }, "easting": { "title": "Easting", "default": -999.0, "type": "number" }, "northing": { "title": "Northing", "default": -999.0, "type": "number" }, "elevation": { "title": "Elevation", "default": -999.0, "type": "number" }, "chans_metadata": { "title": "Chans Metadata", "type": "object", "additionalProperties": { "$ref": "#/definitions/ChanMetadata" } }, "levels_metadata": { "title": "Levels Metadata", "type": "array", "items": { "$ref": "#/definitions/SpectraLevelMetadata" } }, "ref_time": { "title": "Ref Time", "pattern": "%Y-%m-%d %H:%M:%S.%f_%o_%q_%v", "examples": [ "2021-01-01 00:00:00.000061_035156_250000_000000" ] }, "history": { "title": "History", "default": { "records": [] }, "allOf": [ { "$ref": "#/definitions/History" } ] } }, "required": [ "fs", "chans", "n_levels", "first_time", "last_time", "chans_metadata", "levels_metadata", "ref_time" ] } } }
- field tf: resistics.transfunc.TransferFunction [Required]¶
The transfer function that was solved
- field freqs: List[float] [Required]¶
The evaluation frequencies
- field components: Dict[str, resistics.transfunc.Component] [Required]¶
The solution
- field history: resistics.common.History [Required]¶
The processing history
- field contributors: Dict[str, Union[resistics.gather.SiteCombinedMetadata, resistics.spectra.SpectraMetadata]] [Required]¶
The contributors to the solution with their respective details
- property n_freqs¶
Get the number of evaluation frequencies
- property periods: numpy.ndarray¶
Get the periods
- pydantic model resistics.regression.Solver[source]¶
Bases:
resistics.common.ResisticsProcessGeneral resistics solver
Show JSON schema
{ "title": "Solver", "description": "General resistics solver", "type": "object", "properties": { "name": { "title": "Name", "type": "string" } } }
- run(regression_input: resistics.regression.RegressionInputData) resistics.regression.Solution[source]¶
Every solver should have a run method
- field name: Optional[str] [Required]¶
- Validated by
validate_name
- pydantic model resistics.regression.SolverScikit[source]¶
Bases:
resistics.regression.SolverBase class for Scikit learn solvers
Show JSON schema
{ "title": "SolverScikit", "description": "Base class for Scikit learn solvers", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "fit_intercept": { "title": "Fit Intercept", "default": false, "type": "boolean" }, "normalize": { "title": "Normalize", "default": false, "type": "boolean" } } }
- field fit_intercept: bool = False¶
Flag for adding an intercept term
- field normalize: bool = False¶
Flag for normalizing, only used if fit_intercept is True
- pydantic model resistics.regression.SolverScikitOLS[source]¶
Bases:
resistics.regression.SolverScikitOrdinary least squares solver
This is simply a wrapper around the scikit learn least squares regression https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html
Show JSON schema
{ "title": "SolverScikitOLS", "description": "Ordinary least squares solver\n\nThis is simply a wrapper around the scikit learn least squares regression\nhttps://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "fit_intercept": { "title": "Fit Intercept", "default": false, "type": "boolean" }, "normalize": { "title": "Normalize", "default": false, "type": "boolean" }, "n_jobs": { "title": "N Jobs", "default": -2, "type": "integer" } } }
- field n_jobs: int = -2¶
Number of jobs to run
- run(regression_input: resistics.regression.RegressionInputData) resistics.regression.Solution[source]¶
Run ordinary least squares regression on the RegressionInputData
- pydantic model resistics.regression.SolverScikitHuber[source]¶
Bases:
resistics.regression.SolverScikitScikit Huber solver
This is simply a wrapper around the scikit learn Huber Regressor. For more information, please see https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.HuberRegressor.html
Show JSON schema
{ "title": "SolverScikitHuber", "description": "Scikit Huber solver\n\nThis is simply a wrapper around the scikit learn Huber Regressor. For\nmore information, please see\nhttps://scikit-learn.org/stable/modules/generated/sklearn.linear_model.HuberRegressor.html", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "fit_intercept": { "title": "Fit Intercept", "default": false, "type": "boolean" }, "normalize": { "title": "Normalize", "default": false, "type": "boolean" }, "epsilon": { "title": "Epsilon", "default": 1, "type": "number" } } }
- field epsilon: float = 1¶
The smaller the epsilon, the more robust it is to outliers.
- run(regression_input: resistics.regression.RegressionInputData) resistics.regression.Solution[source]¶
Run Huber Regressor regression on the RegressionInputData
- pydantic model resistics.regression.SolverScikitTheilSen[source]¶
Bases:
resistics.regression.SolverScikitScikit Theil Sen solver
This is simply a wrapper around the scikit learn Theil Sen Regressor. For more information, please see https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.TheilSenRegressor.html
Show JSON schema
{ "title": "SolverScikitTheilSen", "description": "Scikit Theil Sen solver\n\nThis is simply a wrapper around the scikit learn Theil Sen Regressor. For\nmore information, please see\nhttps://scikit-learn.org/stable/modules/generated/sklearn.linear_model.TheilSenRegressor.html", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "fit_intercept": { "title": "Fit Intercept", "default": false, "type": "boolean" }, "normalize": { "title": "Normalize", "default": false, "type": "boolean" }, "n_jobs": { "title": "N Jobs", "default": -2, "type": "integer" }, "max_subpopulation": { "title": "Max Subpopulation", "default": 2000, "type": "integer" }, "n_subsamples": { "title": "N Subsamples", "type": "number" } } }
- field n_jobs: int = -2¶
Number of jobs to run
- field max_subpopulation: int = 2000¶
Maximum population. Reduce this if the process is taking a long time
- field n_subsamples: Optional[float] = None¶
Number of rows to use for each solution
- run(regression_input: resistics.regression.RegressionInputData) resistics.regression.Solution[source]¶
Run TheilSen regression on the RegressionInputData
- pydantic model resistics.regression.SolverScikitRANSAC[source]¶
Bases:
resistics.regression.SolverScikitRun a RANSAC solver with LinearRegression as Base Estimator
This is a wrapper around the scikit learn RANSAC regressor. More information can be found here https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.RANSACRegressor.html
Show JSON schema
{ "title": "SolverScikitRANSAC", "description": "Run a RANSAC solver with LinearRegression as Base Estimator\n\nThis is a wrapper around the scikit learn RANSAC regressor. More information\ncan be found here\nhttps://scikit-learn.org/stable/modules/generated/sklearn.linear_model.RANSACRegressor.html", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "fit_intercept": { "title": "Fit Intercept", "default": false, "type": "boolean" }, "normalize": { "title": "Normalize", "default": false, "type": "boolean" }, "min_samples": { "title": "Min Samples", "default": 0.8, "type": "number" }, "max_trials": { "title": "Max Trials", "default": 20, "type": "integer" } } }
- field min_samples: float = 0.8¶
Minimum number of samples in each solution as a proportion of total
- field max_trials: int = 20¶
The maximum number of trials to run
- run(regression_input: resistics.regression.RegressionInputData) resistics.regression.Solution[source]¶
Run RANSAC regression on the RegressionInputData
- pydantic model resistics.regression.SolverScikitWLS[source]¶
Bases:
resistics.regression.SolverScikitOLSWeighted least squares solver
Warning
This is homespun and is currently only experimental
This is simply a wrapper around the scikit learn least squares regression using the sample_weight option https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html
Show JSON schema
{ "title": "SolverScikitWLS", "description": "Weighted least squares solver\n\n.. warning::\n\n This is homespun and is currently only experimental\n\nThis is simply a wrapper around the scikit learn least squares regression\nusing the sample_weight option\nhttps://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "fit_intercept": { "title": "Fit Intercept", "default": false, "type": "boolean" }, "normalize": { "title": "Normalize", "default": false, "type": "boolean" }, "n_jobs": { "title": "N Jobs", "default": -2, "type": "integer" }, "n_iter": { "title": "N Iter", "default": 50, "type": "integer" } } }
- field n_jobs: int = -2¶
Number of jobs to run
- field n_iter: int = 50¶
Number of iterations before quitting if residual is not low enough
- bisquare(r: numpy.ndarray, k: float = 4.685) numpy.ndarray[source]¶
Bisquare location weights
- Parameters
r (np.ndarray) – Residuals
k (float, None) – Tuning parameter. If None, a standard value will be used.
- Returns
weights – The robust weights
- Return type
np.ndarray