Merge pull request #36 from BBVA/develop

4.1 version

Dockerfile

@@ -4,4 +4,4 @@ COPY ./ /app
 COPY ./requirements.txt /tmp/
 COPY ./config/timeout.conf /etc/nginx/conf.d/
 
-RUN pip install mxnet==1.4.1 gluonts && apt-get update && apt-get install -y redis-server python3-celery python-celery-common python3-redis && rm -rf /var/lib/apt/lists/* && pip3 install numpy pandas && pip3 install --requirement /tmp/requirements.txt && pip3 install fbprophet &&  chmod -R g=u /etc/passwd /app  && pip install --upgrade pyflux && pip install --upgrade numpy && rm -fr /root/.cache/pip && service redis-server start
+RUN pip install mxnet==1.4.1 gluonts && apt-get update && apt-get install -y redis-server python3-celery python-celery-common python3-redis && rm -rf /var/lib/apt/lists/* && pip3 install numpy pandas && pip3 install --requirement /tmp/requirements.txt && pip3 install fbprophet &&  chmod -R g=u /etc/passwd /app  && pip install --upgrade pyflux && pip install --upgrade numpy && pip install nbeats-keras && pip install -U tensorflow && rm -fr /root/.cache/pip && service redis-server start

README.md

@@ -34,8 +34,8 @@ To use TIMECOP engine you will need to send a univariate or multivariate time se
 #### Secondary Features
 - [x] Delta / Full timesteps upload
 - [x] Train / Use best known model in production envirotnment
-- [x] Python API for easy integration in notebooks / python apps 
-- [x] Backend Database 
+- [x] Python API for easy integration in notebooks / python apps
+- [x] Backend Database
 
 
 
@@ -48,9 +48,11 @@ To use TIMECOP engine you will need to send a univariate or multivariate time se
 - [x] Gluon-ts
 - [x] Holtwinters
 - [x] ARIMA
+- [x] N-beats
+- [x] Temporal Convolutional Network
 - [ ] TBATS
 
-    
+
 #### Multivariate
 - [x] Long Short Term Memory networks
 - [x] vector autoregression
@@ -70,6 +72,63 @@ To use TIMECOP engine you will need to send a univariate or multivariate time se
 - [x] ChangePoints
 - [ ] Confidence interval
 
+## Installation
+To use timecop the easiest way is to deploy the docker image made with the last version. The steps to use the docker images are:
+
+## Install docker
+
+Download timecop docker image:
+```
+docker pull trakuo/timecop:devel
+```
+
+Execute a container with the timecop image:
+
+```
+docker run --name timecoprunning -p 3000:80 trakuo/timecop:devel
+```
+
+Open web test page to test timecop is running
+```
+http://localhost:3000/static/index.html
+```
+
+## pip python package
+A python package to invoke Timecop from any python code, using the RESTful Web Server API for you.
+For more information, go to the [pip module](https://pypi.org/project/timecopts/)
+
+###### UNIVARIATE
+
+```
+import timecopts as tc
+TS=[1,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9]
+
+URL = 'URL_timecop_server'
+output_json = tc.timecop_univariate(URL,name='test_timeseries', TS,  train=True, restart=True)
+
+```
+
+###### MULTIVARIATE
+
+
+```
+import timecopts as tc
+TS_main=[1,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9]
+
+TS_timeseries= []
+temp = {}
+temp['data'] = [1,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9]
+TS_timeseries.append(temp)
+temp2 = {}
+temp2['data'] = [1,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9,2,3,4,5,6,7,8,9]
+TS_timeseries.append(temp2)
+
+URL = 'URL_timecop_server'
+output_json = tc.timecop_multivariate(URL,name='test_timeseries', TS_main,TS_timeseries,  train=True, restart=True)
+```
+
+
+
 ## Web test page
 
 to use timecop there is another option, a web page that you can use locally and test timecop from CSV files before integrating it into any application