Use Excel to define your model parameters.
Project description
Given an excel file with rows similar to the below
variab le |
sce nar io |
module |
distr ibuti on |
param 1 |
par am 2 |
par am 3 |
un it |
star t date |
end date |
CA GR |
ref date |
labe l |
com men t |
sou rce |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
a |
numpy.random |
choic e |
1 |
kg |
01/0 1/20 09 |
01/0 4/20 09 |
0. 10 |
01/0 1/20 09 |
test var 1 |
|||||
b |
numpy.random |
unifo rm |
2 |
4 |
labe l |
|||||||||
c |
numpy.random |
trian gular |
3 |
6 |
10 |
labe l |
||||||||
d |
bottom_up_com parision.sampli ng_core_route rs |
Distr ibuti on |
core_ router s.csv |
J/ Gb |
labe l |
|||||||||
a |
s1 |
numpy.random |
choic e |
2 |
test var 1 |
|||||||||
multip le choice |
numpy.random |
choic e |
1,2,3 |
kg |
01/0 1/20 07 |
01/0 1/20 09 |
test var 1 |
You can run python/ numpy code that references these variables and generates random distributions.
For example, the following will initialise a variable c with a vector of size 2 with random values from a triangular distribution.
np.random.seed(123) data = ParameterLoader.from_excel('test.xlsx', size=2, sheet_index=0) c = data['c'] >>> [ 7.08471918 5.45131111]
Other types of distributions include choice and normal. However you can specify any distribution from numpy that takes up to three parameters to init.
You can also specify a .csv file with samples and an empiricial distribution function is generated and variable values will be sampled from that.
Scenarios
It is possible to define scenarios and have paramter values for a variable change with each scenario.
data = ParameterLoader.from_excel('test.xlsx', size=1, sheet_index=0) res = data['a'][0] assert res == 1. data.select_scenario('s1') res = data['a'][0] assert res == 2.
use data.unselect_scenario() to return to the default value.
Pandas Dataframes
It is possible to define a time frame for distributions and have sample values change over time.
# the time axis of our dataset times = pd.date_range('2009-01-01', '2009-04-01', freq='MS') # the sample axis our dataset samples = 2 dfl = DataSeriesLoader.from_excel('test.xlsx', times, size=samples, sheet_index=0) res = dfl['a'] assert res.loc[[datetime(2009, 1, 1)]][0] == 1 assert np.abs(res.loc[[datetime(2009, 4, 1)]][0] - pow(1.1, 3. / 12)) < 0.00001
Reload
Reloading data sources is useful when underlying excel files change.
times = pd.date_range('2009-01-01', '2009-04-01', freq='MS') samples = 2 data = MultiSourceLoader() data.add_source(ExcelSeriesLoaderDataSource('test.xlsx', times, size=samples, sheet_index=0)) res = data['a'][0] assert res == 1. wb = load_workbook(filename='test.xlsx') ws = wb.worksheets[0] ws['E2'] = 4. wb.save(filename='test.xlsx') data.reload_sources() res = data['a'][0] assert res == 4. wb = load_workbook(filename='test.xlsx') ws = wb.worksheets[0] ws['E2'] = 1. wb.save(filename='test.xlsx') data.reload_sources() data.set_scenario('s1') res = data['a'][0] assert res == 2. data.reset_scenario() res = data['a'][0] assert res == 1.
Metadata
The contents of the rows is also contained in the metadata
# the time axis of our dataset times = pd.date_range('2009-01-01', '2009-04-01', freq='MS') # the sample axis our dataset samples = 3 dfl = DataSeriesLoader.from_excel('test.xlsx', times, size=samples, sheet_index=0) res = dfl['a'] print(res._metadata)
15.5.2015 0.1.1 Renamed class to ParameterLoader 22.5.2015 0.1.2 Add sheet index as parameter to loader 11.1.2016 0.2.2 Added support to generate pandas dataframes, update to python 3 18.4.2016 0.2.7 Added new flag ‘single_var’ to freeze all variables except one to their mean value - use in sensitivity analysis. 19.8.2016 0.3.0 Upgrade to xarray 0.8.1 20.8.2016 0.3.1 Single var mean now analytical for choice, uniform, triangular and normal; trim white space from var names 4.07.2017 0.4.0 Rewrite with new API 4.07.2017 0.4.1 Added XLWings interface to read from Excel 14.09.2017 0.5.0 Delay sampling from data source until __call__ on Parameter. 16.2.2018 0.5.1 Fixed error in generation of random distributions with zero param values
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distributions
Built Distribution
Hashes for excel_modelling_helper-0.5.5-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | dabe410fbf6d346b8841ac79cb1d1c149f728c7567e5c362880a668d66782ec0 |
|
MD5 | e0e30187dea0ed752fe6dddfda1a8e20 |
|
BLAKE2b-256 | 9e39aa42e49a09a8c04ff18f87d73d94e200531857fa58f36a7da7a8693d414e |