Public
Edited
Aug 13, 2024
mdfn `
## Management
The learning and query process must track and externalize
- correlation between data repository version and model repository version
- data model statistics
- entailment model quality statistics
- [cd4ml](https://martinfowler.com/articles/cd4ml.html):
- validate data : internal learning data integrity, online data equivalence
- component integration : result v/s initial dimension correspondence
- model quality : accuracy, precision, recall
- model bias : relative statistics of learning and on-line data
`
Process Flow
A model is generated from a dataset by defining the learing steps (embedding, meta-parameters) and initiating the learning process.
The result is available in the form of a location, which is incorporated into queries as a functional predicate.
${await(featureStore.image())}
featureStore = FileAttachment("feature-store.svg")
FileAttachment("algo-class-hierarchy.dot+svg.svg").image()
Alternative Implementations
<table>
<tr><th>library</th>
<th>classification</th>
<th>regression</th>>
<th>support vector machines</th>
<th>decision trees</th>
<th>ensemble learning</th>
<th>neural networks</th>
</tr>
<tr><td>scicat-learn</td>
<td>SGDClassifier<br/> OneVsOneClassifier<br/> OneVsRestClassifier <br/>
KNeighborsClassifier</td>
<td>LinearRegression </br>SGDRegressor <br/> LogisticRegression</br>PolynomialFeatures </br>
Ridge </br> Lasso</br> ElasticNet</br> CatBoostRegressor</br> GaussianNB</br>
KernelRidge</br> ElasticNet</br> BayesianRidge</td>
<td>SVC</br> LinearSVC</br> SVR</td>
<td>DecisionTreeClassifier</td>
<td>RandomForestClassifier</br >VotingClassifier</br> GradientBoostingClassifier</br>
LGBMClassifier</br> CatBoostClassifier</br> XGBClassifier<br/>
LGBMRegressor</br> CatBoostRegressor</br> XGBRegressor</td>
</tr>
<tr><td><a href='https://github.com/mmaul/clml'>clml</a></td>
</tr>
<tr><td><a href='https://github.com/melisgl/mgl'>mgl</a>: classification and neural networks from gabor melis</td><td/><td/><td/><td/><td/><td>programmatic definition and controlflow</td>
</tr>
</table>
viewof footnotes = footlist([
{ref: 'shashua2008', text: `Introduction to Machine Learning, p1`},
{ref: 'koehrsen2018', text: `Koehrsen [Neural Network Embeddings Explained](https://towardsdatascience.com/neural-network-embeddings-explained-4d028e6f0526)`},
{ref: 'braga2020', text: `Braga, J., Dias, J.L. and Regateiro, F., 2020. [A MACHINE LEARNING ONTOLOGY](https://frenxiv.org/rc954/download/?format=pdf)`}
])
import {mdfn, footlist, footnoteCSS} with { footnotes as footnotes } from '@formsandlines/linked-markdown-style-footnotes-with-tooltips'
footnoteCSS
Purpose-built for displays of data
Observable is your go-to platform for exploring data and creating expressive data visualizations. Use reactive JavaScript notebooks for prototyping and a collaborative canvas for visual data exploration and dashboard creation.
