{"id":34618565,"url":"https://github.com/asdf8601/aguathon_2019","last_synced_at":"2026-05-26T12:07:11.506Z","repository":{"id":77112352,"uuid":"173186010","full_name":"asdf8601/aguathon_2019","owner":"asdf8601","description":"Aguathon 2019 ITAINNOVA River Ebro Hackathon ","archived":false,"fork":false,"pushed_at":"2019-10-03T16:41:19.000Z","size":76190,"stargazers_count":0,"open_issues_count":2,"forks_count":0,"subscribers_count":1,"default_branch":"master","last_synced_at":"2026-02-01T15:56:16.638Z","etag":null,"topics":["ebro","hackathon","river"],"latest_commit_sha":null,"homepage":"","language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/asdf8601.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null}},"created_at":"2019-02-28T20:55:17.000Z","updated_at":"2021-02-02T10:38:27.000Z","dependencies_parsed_at":"2023-02-26T19:15:25.764Z","dependency_job_id":null,"html_url":"https://github.com/asdf8601/aguathon_2019","commit_stats":{"total_commits":57,"total_committers":2,"mean_commits":28.5,"dds":0.01754385964912286,"last_synced_commit":"95bc627e648ef07d2ab6275ee81b4f5b05307f20"},"previous_names":["asdf8601/aguathon_2019","mmngreco/aguathon_2019"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/asdf8601/aguathon_2019","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/asdf8601%2Faguathon_2019","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/asdf8601%2Faguathon_2019/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/asdf8601%2Faguathon_2019/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/asdf8601%2Faguathon_2019/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/asdf8601","download_url":"https://codeload.github.com/asdf8601/aguathon_2019/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/asdf8601%2Faguathon_2019/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":33519267,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T03:12:49.672Z","status":"ssl_error","status_checked_at":"2026-05-26T03:12:47.976Z","response_time":63,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ebro","hackathon","river"],"created_at":"2025-12-24T14:56:52.450Z","updated_at":"2026-05-26T12:07:11.500Z","avatar_url":"https://github.com/asdf8601.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Aguathon 2019\n\n\n| Autor  | Maximiliano Greco            |\n| :---:  | :---------------:            |\n| Github | https://github.com/mmngreco/ |\n\n\n\n## 1. Instalación\n\n### Conda\n\nDefinir un entorno virtual con conda.\n\n```bash\nconda env create -f environment.yml -n river\nconda activate river\n```\n\n### Docker\n\n```bash\ndocker build -t aguathon .\ndocker run -it aguathon\n```\n\n## 2. Uso\n\nPara entrenar modelos:\n\n```python\npython CNN_feat2d.py -h  # para entrenar (no es necesario)\n```\n\nPara predecir con los modelos entrenados, solo hay que ejecutar:\n\n```python\npython predict.py  # para predecir\n```\n\n\n## 3. Motivación\n\nPara la prediccón se ha usado una red neuronal que mezcla una capa\n\"convolucional\", seguida de una capa oculta  y una capa \"full conected\" de\nsalida. Esta arquitectura es la que mejor me ha funcionado, otros proyectos que\nhe visto, usan LSTM o GRU, sin embargo, esa aproximación no me ha dado buenos\nresultados con el añadido de que poco intuitivas y dificiles de entender.\n\nEsta estructura la he definido a partir de la intuición y del ensayo error.\nLa mayoria de trabajos y proyectos que he consulado no hacían uso de las redes\nrecurrentes, en su lugar aplicaban redes neuronals o bien LSTM. Me pareció que\nmerecía la pena hacer la prueba ya que estas redes se aplican en imagen\nfundamentalmente. En nuestro caso, a pesar de tener series temporales, parece\nque funcionan especialmente bien debido a la similitud dimensional con las\nimagenes, en cierta, forma estos río forman una \"imagen\" que queremos predecir.\n\n\n\n### Estructura de la RED\n\nPuede ver este diagrama usando un renderizador de [`mermaid`](https://mermaidjs.github.io/mermaid-live-editor/#/view/eyJjb2RlIjoiZ3JhcGggTFJcblhfMjQoKFgpKSAtLT4gQ05OXzFkXzI0XG55MjQoKHkyNCkpIC0tPiBDTk5fMWRfMjRcbkNOTl8xZF8yNCAtLT4gRGVuc2UxXzI0XG5EZW5zZTFfMjQgLS0-IERlbnNlMl8yNFxuRGVuc2UyXzI0IC0tPiBGbGF0dGVuXzI0XG5GbGF0dGVuXzI0IC0tPiBEZW5zZTNfMjRcbkRlbnNlM18yNCAtLT4geWhhdF8yNCgoeWhhdDI0KSlcblxuXG5YXzQ4KChYKSkgLS0-IENOTl8xZF80OFxueV80OCgoeTQ4KSkgLS0-IENOTl8xZF80OFxuQ05OXzFkXzQ4IC0tPiBEZW5zZTFfNDhcbkRlbnNlMV80OCAtLT4gRGVuc2UyXzQ4XG5EZW5zZTJfNDggLS0-IEZsYXR0ZW5fNDhcbkZsYXR0ZW5fNDggLS0-IERlbnNlM180OFxuRGVuc2UzXzQ4IC0tPiBCXzQ4KCh5aGF0NDgpKVxuXG5YXzcyKChYKSkgLS0-IENOTl8xZF83MlxueV83MigoeTcyKSkgLS0-IENOTl8xZF83MlxuQ05OXzFkXzcyIC0tPiBEZW5zZTFfNzJcbkRlbnNlMV83MiAtLT4gRGVuc2UyXzcyXG5EZW5zZTJfNzIgLS0-IEZsYXR0ZW5fNzJcbkZsYXR0ZW5fNzIgLS0-IERlbnNlM183MlxuRGVuc2UzXzcyIC0tPiBCXzcyKCh5aGF0NzIpKSIsIm1lcm1haWQiOnsidGhlbWUiOiJkZWZhdWx0In19).\n\n![graph](./assets/figures/mermaid-diagram-20190501164854.svg)\n\nEn el script `CNN_feat2d.py` se encuentra la función `build_model()` que se encarga\nde crear la red de acuerdo a unos parámetros. Pero en esencia es lo siguiente:\n\n```python\nmodel = Sequential()\nmodel.add(Conv1D(\n    filters=filters,\n    kernel_size=kernel_size,\n    activation='relu',\n    input_shape=(n_steps, n_features),\n))\nmodel.add(Dense(\n    NEURONS1,\n    activation='relu',\n    kernel_regularizer=L1L2,\n))\nmodel.add(Dense(\n    n_features,\n    activation='relu',\n    kernel_regularizer=L1L2,\n))\nmodel.add(Flatten())\nmodel.add(Dense(NEURONS_OUT))\nmodel.compile(optimizer='adam', loss='mse')\n```\n\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fasdf8601%2Faguathon_2019","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fasdf8601%2Faguathon_2019","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fasdf8601%2Faguathon_2019/lists"}