52f40c400ad2c02336eb788bac77f1e55d16c4bb
[talweg.git] / README.md
1 # Predict PM10 as functions of time
2
3 Joint work with [Jean-Michel Poggi](http://www.math.u-psud.fr/~poggi/) and [Bruno Portier](http://lmi2.insa-rouen.fr/~bportier/)
4
5 ---
6
7 Forecast a curve sampled within the day (seconds, minutes, hours...),
8 using past measured curves + past exogenous informations, which could be some aggregated
9 measure on the past curves, the weather... Main starting point: computeForecast().
10
11 NOTE: algorithms are not specific to PM10 and could be applied to anything else (in similar contexts).
12 However, seasons are hard-coded to follow pollution events.
13 You may want to change them in pkg/R/utils.R, function .isSameSeason()
14
15 NOTE 2: the package works for only one series (one station). Further extensions might consider the multi-series case.
16
17 ---
18
19 The final report may be found at [this location](http://www.airnormand.fr/Publications/Publications-telechargeables/Rapports-d-etudes)