0.11.0

What's Changed

Major Changes 🛠

• Bump pymc dependency by @ricardoV94 in #1269
• Budget optimizer refactor by @ricardoV94 in #1357

New Features 🎉

• add hdi_list kwarg to plot_posterior_predictive by @drbenvincent in #1096
• Allow Negative Values in Media Channels by @louismagowan in #1092
• Add plot_expected_purchases_over_time function by @ColtAllen in #1115
• Beta geo expected probability no purchase by @DylanZammit in #1094
• Harmonize prior and posterior predictive plots in MMM module by @juanitorduz in #1170
• Create load_from_idata method by @wd60622 in #1211
• Add CLV plot_expected_purchases_ppc by @ColtAllen in #1222
• Include tvp scaling in contribution grid by @wd60622 in #1253
• Make optimization based on the entire posterior and not on the marginal mean parameters. by @cetagostini in #1151
• HSGP as component by @wd60622 in #1246
• Add Gradient-Free Fit Method to CLV Models by @ColtAllen in #1266
• Add graphviz method to ModelBuilder class by @ColtAllen in #1284
• Improved Default Priors for BetaGeoModel by @ColtAllen in #1264
• Non-centered ZeroSumNormal Prior support by @wd60622 in #1290
• Allow non-Prior parameters in Prior and media transformations by @wd60622 in #1292
• Deserialize arbitrary objects by @wd60622 in #1296
• Modified Beta Geometric model by @PabloRoque in #1301
• Allow VariableFactory in FourierBase by @aseyboldt in #1304
• Add Censored wrapper for Prior class by @wd60622 in #1309
• Register new media transformations automatically by @wd60622 in #1320
• Add new customer_choice module by @drbenvincent in #803
• 1314 implement serialization for fourier classes by @wd60622 in #1315
• Deprecate to/from_json of Prior class in favor of to/from_dict by @wd60622 in #1338
• Add autologging for CLV models by @wd60622 in #1354
• Introduce BetaGeoNBD Random Variable by @PabloRoque in #1317
• MMM Evaluation, Diagnostics and MLFlow Registry by @louismagowan in #1368
• MLflow logging before slow function calls by @wd60622 in #1393
• Custom optimizer constraints by @cetagostini in #1358
• Creating Causal Identification module by @cetagostini in #1166

Bugfixes 🐛

• Fix LinearTrend rending issue by @takechanman1228 in #1173
• Fix bug in the "Causal Ladder" docs page by @drbenvincent in #1175
• Solving bug on optimization plot by @cetagostini in #1174
• patch values stacked plot by @juanitorduz in #1181
• [FIX] Wrong legend in the posterior predictive plot then we set gradient=True by @juanitorduz in #1212
• Fix budget allocation plot in case study by @juanitorduz in #1216
• Raise error when two dims are not compatible by @wd60622 in #1249
• 1271 predictions=true for sample posterior predictive throws error by @wd60622 in #1272
• Fix tanh component by @aseyboldt in #1303
• Enable transform _sample_curve to work with VariableFactory by @lucianopaz in #1362
• keep date column in test_scale_lift_measurements by @malitsadok1 in #1316
• Moved pyprojroot from docs to dependencies by @nayan2167 in #1417
• Remove sample_posterior_predictive_patch by @wd60622 in #1414

Documentation 📖

• Expose Streamlit App in Documentation by @louismagowan in #1091
• Fix typos in MMM Example notebook by @wd60622 in #1146
• add mention of GPU in documentation by @wd60622 in #1152
• Update MMM docs to use graphviz method by @takechanman1228 in #1172
• MMM Case Study Fixes by @juanitorduz in #1169
• Run sBG with seed by @wd60622 in #1189
• Fix beta ref mmm by @juanitorduz in #1194
• Fix minor typos in mmm example notebook by @SangamSwadiK in #1215
• CLV Quickstart Revisions by @ColtAllen in #1243
• fix typo and change first lines mood by @wd60622 in #1254
• Link sBG Notebook in README and Index by @ColtAllen in #1265
• Update README.md (fix badges) by @juanitorduz in #1278
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in #1313
• customer choice: improve plots and harmonize colors by @juanitorduz in #1322
• Re-run TVP mmm notebook by @juanitorduz in #1324
• add customer choice to readme by @juanitorduz in #1326
• sort docs sections by @juanitorduz in #1327
• Generalize verbiage of installation instructions by @wd60622 in #1335
• Prior predictive guide by @juanitorduz in #1334
• hot fix: budget allowcation notebooks by @juanitorduz in #1389
• Breakout and Document MLflow functions by @wd60622 in #1396
• Relax the type that is needed for evaluation metric function by @wd60622 in #1413
• Add CLV Roadmap by @ColtAllen in #1364
• CLV Documentation Revisions by @ColtAllen in #1419

Maintenance 🔧

• Remove duplicate attrs/functions from MMM by @sreekailash in #1336
• Move fit_result to model_builder.py and remove redundancies from CLV and MMM by @sreekailash in #1344
• Remove media transformation registration functions by @wd60622 in #1378
• Add load_from_idata method by @wd60622 in #1379
• Add tests by @juanitorduz in #1391
• Implement ModifiedBetaGeoNBD and ModifiedBetaGeoNBDRV by @PabloRoque in #1375

New Contributors

• @nhan295 made their first contribution in #1114
• @DylanZammit made their first contribution in #1094
• @dependabot made their first contribution in #1134
• @nataziel made their first contribution in #1285
• @aseyboldt made their first contribution in #1303
• @sreekailash made their first contribution in #1336
• @malitsadok1 made their first contribution in #1316
• @nayan2167 made their first contribution in #1412

Full Changelog: 0.10.0...0.11.0

0.10.0

What's Changed

New Features 🎉

• #1063 Pareto_nbd distribution enhancements by @Ishaanjolly in #1067
• #990 - Plot FourierBase along date rather than index by @Ishaanjolly in #1068
• Add CLV Cumulative Transaction Utlity by @ColtAllen in #1076

Bugfixes 🐛

• michaelis_menten transformation as pt.TensorVariable by @PabloRoque in #1054

Documentation 📖

• comment save notebook in mmm example by @juanitorduz in #1053
• #1020 Correction of doctring for plot.py for sample_kwargs by @Ishaanjolly in #1060
• remove spend scaling step from budget optimizer by @juanitorduz in #1070
• added plotting options to plot_posterior_predictive by @jsnyde0 in #1072
• Docs update: make sure the right title shows up on Google by @ulfaslak in #1074
• small fixes mm roas notebook by @juanitorduz in #1083
• New notebook: causal inference with MMM's by @drbenvincent in #1032
• Polishing the new new causal ladder docs page by @drbenvincent in #1084
• Fix malformed equation in MMM docstring by @wd60622 in #1085
• MMM Case Study from PyData Global by @juanitorduz in #1044

Maintenance 🔧

• #1035 Distribution new customer enhancements by @Ishaanjolly in #1061
• Deprecate several pymc_marketing.mmm.utils functions by @PabloRoque in #1078
• Automatic labeling of untouched issues by @wd60622 in #1086

New Contributors

• @jsnyde0 made their first contribution in #1072

Full Changelog: 0.9.0...0.10.0

0.9.0

What's Changed

New Features 🎉

• Add user customization to plot_curve methods by @wd60622 in #1018
• Add BetaGeoBetaBinomModel by @ColtAllen in #1031
• MLFlow auto logging by @wd60622 #921

Documentation 📖

• Enhance MMM intro material by @juanitorduz in #1017
• Improve landing page by @juanitorduz in #1023
• simplify dag from mmm notebook by @juanitorduz in #1046

Maintenance 🔧

• lower bound xarray by @juanitorduz in #1025
• Mute Sampling: [] stdout when calling get_channel_contributions_forward_pass_grid by @PabloRoque in #1043

Full Changelog: 0.8.0...0.9.0

0.8.0

What's Changed

New Features 🎉

• Prior class to represent distributions by @wd60622 in #759
• Pull out seasonality as YearlyFourier and MonthlyFourier by @wd60622 in #802
• Separate Weibull adstock into CDF & PDF by @wd60622 in #810
• PoC: Use Pydantic as data validator by @juanitorduz in #809
• Create inverse_scaled_logistic_saturation and the corresponding class by @arthurmello in #827
• use labs theme as external by @OriolAbril in #830
• Skip coords with scalar value by @GiannisApost in #868
• Allow plot MMM components in the original scale by @juanitorduz in #870
• Add root saturation function (issue #702) by @iraur in #858

Bugfixes 🐛

• Save & load support for time varying parameters by @wd60622 in #815
• json.loads with python types bug by @wd60622 in #881
• Save off media transformations by @wd60622 in #882

Documentation 📖

• add intercept and target variable to MMM component example by @wd60622 in #850
• Various MMM small documentation fixes by @juanitorduz in #854
• Update UML diagrams by @drbenvincent in #856
• Fix model_builder docstrings by @GiannisApost in #861
• add GH discussions link to README by @cluhmann in #866
• Fix Visual for hill_saturation function (Issue #851 ) by @PatrickRobotham in #857
• Inference changed to dataset by @Ishaanjolly in #873
• Fix default ConvMode in docstring by @ferrine in #864
• Update deployment docs by @wd60622 in #887

Maintenance 🔧

• Make PyMC 5.13.0 the minimum version by @wd60622 in #816
• Date Validation and MMM Model Hamonization (Pydantic) by @juanitorduz in #824
• Future-proof prior_linearized method call by @shuvayan in #806
• Specify 0.9.0 as deprecation version for recent MMM changes by @wd60622 in #849
• Fixing ruff commands in Makefile #825 by @MuradKhalil in #859
• Check for missing attrs after sample_prior_predictive and fit by @wd60622 in #867

New Contributors

• @shuvayan made their first contribution in #806
• @arthurmello made their first contribution in #827
• @MuradKhalil made their first contribution in #859
• @GiannisApost made their first contribution in #861
• @PatrickRobotham made their first contribution in #857
• @Ishaanjolly made their first contribution in #873
• @iraur made their first contribution in #858

Full Changelog: 0.7.0...0.8.0

0.7.0

What's Changed

This release comes with great new features and improvements! Among many things, we have:

• Internal MMM refactor to make it customizable!
  • We allow custom adstock and saturation functions.
  • MMM time-varying media-baseline component available!
  • Hierarchical distributions.
• Streamlit explainer app! See https://pymc-marketing-app.streamlit.app/
• CLV API standardization
• RFM segmentation
• Documentation improvements and new examples!

Major Changes 🛠

• Update BetaGeoModel API by @ColtAllen in #709
• User-defined media transformations and custom ordering by @cetagostini in #632
• CLV Plotting API by @ColtAllen in #728
• GammaGammaModel Docstrings and API Standardization by @ColtAllen in #758

New Features 🎉

• v0 Streamlit MMM Explainer App by @louismagowan in #614
• Add BetaGeoBetaBinom Distribution Block by @ColtAllen in #431
• RFM Segmentation by @ColtAllen in #680
• Media transformation sampling & plotting methods by @wd60622 in #734
• Hierarchical Model Configuration by @wd60622 in #743
• Allowing Hierarchical Non Centered Parametrization by @cetagostini in #747
• Creating Time Base component for Media Contribution by @cetagostini in #752

Bugfixes 🐛

• Fix related to column renaming after aggregating test frequency by @IvanUgrin in #698
• model.fit doesn't remove prior samples by @wd60622 in #741
• Deepcopy of posterior to allow second fit call by @wd60622 in #790
• Add prior predictive example notebook by @juanitorduz in #787

Documentation 📖

• Correct BetaGeo docstring example by @wd60622 in #693
• Fix some ParetoNBDModel docstring typos by @Mews in #731
• Minor improvements [MMM] by @juanitorduz in #735
• MMM Component Notebook by @wd60622 in #748
• start addressing sphinx warnings and rendering issues by @OriolAbril in #750
• CLV Modeling Domains and Docstrings by @ColtAllen in #785
• Fix TVP Notebook by @juanitorduz in #793
• docs: Update model_builder.py to resolve warning in documentation build by @c0d33ngr in #797
• Time Varying Media Contribution Notebook by @cetagostini in #778

Maintenance 🔧

• pass kwargs to minimizer by @juanitorduz in #737
• improve tests mmm utils by @juanitorduz in #738
• fix np typing by @juanitorduz in #763
• remove noqa from plots by @juanitorduz in #761
• Run Ruff Notebooks by @juanitorduz in #773
• Run example notebooks CI by @juanitorduz in #791

New Contributors

• @louismagowan made their first contribution in #614
• @IvanUgrin made their first contribution in #698
• @Mews made their first contribution in #731
• @c0d33ngr made their first contribution in #797

Full Changelog: 0.6.0...0.7.0

0.6.0

What's Changed

New Features 🎉

• Time-varying intercept by @ulfaslak in #628
• New Pareto/NBD Example Notebook by @ColtAllen in #646
• MMM Example Notebook Improvements (waterfall & error plots) by @juanitorduz in #664
• Plot Waterfall Components Decomposition by @cetagostini-wise in #631

Bugfixes 🐛

• Pass conv mode to adstock functions by @juanitorduz in #665
• Fix compressed images in docs. by @juanitorduz in #667
• Fix ylabel BG/NBD Notebook by @juanitorduz in #654
• TV intercept notebook: fix title level by @juanitorduz in #663
• Add sample_kwargs BG/NBD Notebook by @juanitorduz in #676

Documentation 📖

• Update resources.md by @coreyabs-db in #652
• Improve tanh saturation parameter class documentation by @wd60622 in #657
• Add quickstart to readme by @juanitorduz in #653
• Add contributors to README by @juanitorduz in #659
• Update README.md by @nialloulton in #660

New Contributors

• @coreyabs-db made their first contribution in #652

Full Changelog: 0.5.0...0.6.0

0.5.0

What's Changed

New Features 🎉

• Add rfm_train_test_split Utility Function by @ColtAllen in #572

Bugfixes 🐛

• Fix clv plotting bugs and edits to Quickstart by @ColtAllen in #601

Documentation 📖

• Add lift tests check to comparison table by @juanitorduz in #611
• Add more content to the Gamma-Gamma Notebook by @juanitorduz in #573
• Add more content to the BG/NBD Notebook by @juanitorduz in #571
• Improve MMM Docs by @juanitorduz in #612
• MMM notebook small fixes by @juanitorduz in #626

Maintenance 🔧

• Drop Python 3.9 support by @juanitorduz in #615
• Add support for Python 3.12 by @juanitorduz in #618

0.4.2

What's Changed

Bugfixes 🐛

• Fix bug in predictive methods of ParetoNBD model with covariates by @ricardoV94 in #589

Full Changelog: 0.4.1...0.4.2

0.4.1

What's Changed

New Features 🎉

• Implement ParetoNBD with covariates by @ricardoV94 in #545
• Compute channel contributions from new spends by @wd60622 in #456
• Add Docker Image by @juanitorduz in #540

Bugfixes 🐛

• Fix repr error when CLVModel is not yet built by @ricardoV94 in #563

Documentation 📖

• Update resources page by @juanitorduz in #539
• Add to documentation feature comparison by @wd60622 in #550
• Make BetaGeoModel docstring example work by @wd60622 in #558
• Add nbqa-ruff for doc notebooks by @juanitorduz in #561
• Add other nuts samplers example notebook by @juanitorduz in #579

Maintenance 🔧

• Use union of default and user provided configs in ModelBuilder by @ricardoV94 in #565
• Remove scale_preserving_logistic_saturation function by @ulfaslak in #585

New Contributors

• @AlexAndorra made their first contribution in #556

Full Changelog: 0.4.0...0.4.1

0.4.0

What's Changed

New Features 🎉

• Handle new data correctly and extend the functionality of MMM posterior predictive methods by @wd60622 in #482
• Add baselined saturation by @ferrine in #498
• Add scale preserving saturation function by @ulfaslak in #503
• Add Weibull CDF and PDF Adstock Transformations by @abdalazizrashid in #499

Bugfixes 🐛

• Swap Before and After convolution modes by @abdalazizrashid in #501
• Support use of ParetoNBDModel in customer_lifetime_value by @ColtAllen in #512

Documentation 📖

• Add notebook section for out-of-sample predictions by @wd60622 in #504

New Contributors

• @ferrine made their first contribution in #498
• @ulfaslak made their first contribution in #503

Full Changelog: 0.3.3...0.4.0