release v0.1.5 - "ensemble tolerance"

debian/changelog

@@ -1,3 +1,9 @@
+simbuto (0.1.5) unstable; urgency=medium
+
+  * implement ensemble tolerance forecast
+
+ -- Yann Büchau <[email protected]>  Fri, 13 Jan 2017 22:48:00 +0100
+
 simbuto (0.1.4) unstable; urgency=medium
 
   * implement amount and day tolerance

usr/lib/simbuto/python/simbuto/__init__.py

@@ -6,7 +6,7 @@
 # Internal modules
 
 # the version
-VERSION = "0.1.4"
+VERSION = "0.1.5"
 
 __version__ = VERSION
 

usr/lib/simbuto/python/simbuto/gui.py

@@ -37,6 +37,9 @@ def __init__(self):
             signals = [signal.SIGINT, signal.SIGTERM, signal.SIGHUP],
             handler = self.quit
         )
+        # can't use Gtk.main() because of a bug that prevents proper SIGINT
+        # handling. use Glib.MainLoop() directly instead.
+        self.mainloop = GLib.MainLoop() # main loop
 
 
     ##################
@@ -74,6 +77,10 @@ def currently_edited_file(self):
         except AttributeError:
             return None
 
+    @property
+    def is_running(self):
+        return self.mainloop.is_running()
+
     @currently_edited_file.setter
     def currently_edited_file(self, value):
         self._currently_edited_file = value
@@ -405,6 +412,15 @@ def setup_gui(self):
         # statusbar
         self.reset_statusbar() # initially reset statusbar
 
+        # settings
+        self("ensemble_expander_label").set_text(_("Statistics"))
+        self("ensemble_settings_useensemble_checkbutton").set_label(_(
+            "display ensemble"))
+        self("ensemble_settings_useensemble_checkbutton").set_tooltip_text(_( 
+            "[slower] Based on the given day and amount tolerances, run an " 
+            "ensemble and display the 10% and 90% quantiles as darker shadow."))
+
+
         # calendar
         # pretend the start date was selected and let automatic range selection
         # do the rest
@@ -465,40 +481,51 @@ def update_statusbar(self, text = None):
         statuslabel.set_text(newtext)
 
     def update_graph_from_editor(self, *args, size=None):
-        # rect = self("plot_image").get_allocation()
-        if size is None: # use current size
-            rect = self("plot_scrolledwindow").get_allocation()
-            width = rect.width
-            height = rect.height
-        else: # use given size
-            width, height = size
-
-        try:
-            currentfile = os.path.basename(self.currently_edited_file)
-        except AttributeError:
-            currentfile = _("unnamed-budget")
-        name =  "{}.png".format(currentfile)
-        filename = os.path.join(config.personal_simbuto_dotfolder(),
-            "plots",name)
-        self.update_statusbar(_("updating graph..."))
-        success = self.signalmanager.emit_signal("create-graph-from-text",
-            filename=filename, # to this file
-            text = self.current_editor_content, # this text
-            width = width, height = height, # these dimensions
-            start = self.selected_start_date, # this start date
-            end = self.selected_end_date, # this end date
-            )
-        if success[0]:
-            self.logger.debug(_("The graph file was obviously " 
-                "sucessfully updated."))
-            self.update_graph_from_file(filename)
-            self.update_statusbar(_("Graph updated"))
+        if self.is_running: # only if gui is running
+            # rect = self("plot_image").get_allocation()
+            if size is None: # use current size
+                rect = self("plot_scrolledwindow").get_allocation()
+                width = rect.width
+                height = rect.height
+            else: # use given size
+                width, height = size
+
+            try:
+                currentfile = os.path.basename(self.currently_edited_file)
+            except AttributeError:
+                currentfile = _("unnamed-budget")
+
+            cb = self("ensemble_settings_useensemble_checkbutton")
+            use_ensemble = cb.get_active()
+
+            name =  "{}.png".format(currentfile)
+            filename = os.path.join(config.personal_simbuto_dotfolder(),
+                "plots",name)
+            self.update_statusbar(_("updating graph..."))
+            success = self.signalmanager.emit_signal("create-graph-from-text",
+                filename=filename, # to this file
+                text = self.current_editor_content, # this text
+                width = width, height = height, # these dimensions
+                start = self.selected_start_date, # this start date
+                end = self.selected_end_date, # this end date
+                use_ensemble = use_ensemble, # use the ensemble or not
+                )
+            if success[0]:
+                self.logger.debug(_("The graph file was obviously " 
+                    "sucessfully updated."))
+                self.update_graph_from_file(filename)
+                self.update_statusbar(_("Graph updated"))
+            else:
+                self.logger.debug(_("There was a problem updating the graph."))
+                self.update_statusbar(_("[WARNING] There was a problem " 
+                    "updating the graph. Please check the input!"))
+            return True
         else:
-            self.logger.debug(_("There was a problem updating the graph."))
-            self.update_statusbar(_("[WARNING] There was a problem " 
-                "updating the graph. Please check the input!"))
-
-        return True
+            self.updating_graph_from_editor_is_now_okay = True
+            self.logger.debug(_("The gui is not running. Refusing to update " 
+                "the graph now! Remembering this for gui start."))
+            return False
+
 
     def update_graph_from_file(self, filename):
         self("plot_image").set_from_file(filename)
@@ -747,9 +774,6 @@ def fill_editor_from_file(self, filename):
 
     # run the gui
     def run(self):
-        # can't use Gtk.main() because of a bug that prevents proper SIGINT
-        # handling. use Glib.MainLoop() directly instead.
-        self.mainloop = GLib.MainLoop() # main loop
         # signal.signal(signal.SIGINT, signal.SIG_DFL)
         self.logger.debug(_("Starting GLib main loop..."))
         self.mainloop.run()

usr/lib/simbuto/python/simbuto/manager.py

@@ -105,7 +105,9 @@ def md5sum_of_file(self, filename):
     def create_png_graph_from_text(self, text, filename, 
         width = 600, height = 400, 
         start = datetime.datetime.now(), 
-        end = datetime.datetime.now() + datetime.timedelta(365) ):
+        end = datetime.datetime.now() + datetime.timedelta(365),
+        ensemble_size = 100,
+        use_ensemble = False):
         """ Create a png graph from simbuto csv-like text
         Args:
             text (str): the csv-like simbuto budget
@@ -117,21 +119,27 @@ def create_png_graph_from_text(self, text, filename,
             end [Optional(datetime.datetime)]: the end time of the budget
                 calculation and plotting. Defaults to the current day plus one
                 year.
+            use_ensemble [Optional(bool)]: calculat an ensemble? Defaults to
+                False.
+            ensemble_size [Optional(int)]: The ensemble size to use. Defaults to 
+                100.
         Returns:
             success (bool): True if graph png file was created, False otherwise
         """
         start_date = R("as.Date('{}-{}-{}')".format(
             start.year,start.month,start.day))
         end_date = R("as.Date('{}-{}-{}')".format(
             end.year,end.month,end.day))
+        if not use_ensemble:
+            ensemble_size = R("NULL")
         try:
             # append newline
             if not text.endswith("\n"): text += "\n"
             # create the budget from text
             budget_frame = R.read_budget_from_text(text = text)
             # create the timeseries from the budget
             timeseries_frame = R.timeseries_from_budget(budget = budget_frame,
-                start = start_date, end = end_date)
+                start = start_date, end = end_date, ensemble_size=ensemble_size)
             # plot to png
             R.plot_budget_timeseries_to_png(filename=filename,
                 timeseries = timeseries_frame, width = width, height = height)

usr/lib/simbuto/r/simbuto-functions.R

@@ -24,9 +24,10 @@ timeseries_from_budget <- function(
     # create empty frame with day series
     all.days <- seq.Date(from = start, to = end, by = "days")
     MONEY <- data.frame(day = all.days, amount = 0)
+    N <- nrow(MONEY)
 
     # start with empty series
-    worstcase <- bestcase <- undisturbed <- rep(0, nrow(MONEY))
+    worstcase <- bestcase <- undisturbed <- rep(0, N)
     # loop over all facts
     for (factnr in 1:nrow(budget)) {
         fact <- budget[factnr,] # current fact
@@ -39,7 +40,7 @@ timeseries_from_budget <- function(
             fact.end <- fact.start
             interval = "day" # pick any interval, doesn't matter
         }
-        # cat("from=",fact.start," to=",fact.end," by=",interval," length.out=",number.occurences,"\n")
+        # print(fact)
         occurences <- c()
         if(fact.start <= fact.end) {
             occurences <- seq.Date(from = fact.start, to = fact.end, by = interval)
@@ -49,20 +50,54 @@ timeseries_from_budget <- function(
 
         # get the indices
         indices <- na.omit(match(x = occurences, table = MONEY$day))
+
+        # create the series
+        # undisturbed - original
         undisturbed <- undisturbed + fact_amounts_series(
             occurences = occurences_bool, fact = fact,with_tolerance = FALSE)
+        # worst case
         worstcase <- worstcase + fact_amounts_series(
             occurences = occurences_bool, fact = fact,with_tolerance = TRUE, 
             worst_case = TRUE )
+        # best case
         bestcase <- bestcase + fact_amounts_series(
             occurences = occurences_bool, fact = fact,with_tolerance = TRUE, 
             worst_case = FALSE )
+        # ensemble
+        if(any(is.finite(ensemble_size))) {
+            if(!exists("ensemble")) # create variable if not existant yet
+                ensemble <- matrix(0,nrow=ensemble_size, ncol = N)
+            # do the runs
+            # cat("create members...")
+            for(i in 1:ensemble_size) {
+                tendencies <- fact_amounts_series(
+                    occurences = occurences_bool, fact = fact,with_tolerance = TRUE, 
+                    random_tolerance = TRUE)
+                # add to ensemble run
+                ensemble[i,] <- ensemble[i,] + tendencies
+            }
+        }
     }
 
     # cumulate
     MONEY$amount    = cumsum(undisturbed)
     MONEY$worstcase = cumsum(worstcase)
     MONEY$bestcase  = cumsum(bestcase)
+    # ensemble
+    if(exists("ensemble")) {
+        # cumulate
+        ensemble <- t(apply(X = ensemble, MARGIN = 1, FUN = cumsum))
+        # MONEY$ensmean <- apply(X = ensemble,MARGIN = 2, FUN = mean)
+        # MONEY$ensmedian <- apply(X = ensemble,MARGIN = 2, FUN = median)
+        # MONEY$ensquant25 <- apply(X = ensemble,MARGIN = 2, FUN = function(x)quantile(x,probs = c(0.25)))
+        # MONEY$ensquant75 <- apply(X = ensemble,MARGIN = 2, FUN = function(x)quantile(x,probs = c(0.75)))
+        # MONEY$ensquant10 <- apply(X = ensemble,MARGIN = 2, FUN = function(x)quantile(x,probs = c(0.10)))
+        # MONEY$ensquant90 <- apply(X = ensemble,MARGIN = 2, FUN = function(x)quantile(x,probs = c(0.90)))
+        MONEY$ensquant05 <- apply(X = ensemble,MARGIN = 2, FUN = function(x)quantile(x,probs = c(0.05)))
+        MONEY$ensquant95 <- apply(X = ensemble,MARGIN = 2, FUN = function(x)quantile(x,probs = c(0.95)))
+        # MONEY$ensmin <- apply(X = ensemble,MARGIN = 2, FUN = min)
+        # MONEY$ensmax <- apply(X = ensemble,MARGIN = 2, FUN = max)
+    }
     # empty data frame
     return(MONEY)
 }
@@ -98,15 +133,15 @@ fact_amounts_series <- function(
     if(with_tolerance) {
         if(random_tolerance) {
             # modify amount randomly
-            amounts <- runif( n = length(indices), 
+            amounts <- round(runif( n = length(indices), 
                               min = fact_amount - fact_tolerance_amount,
                               max = fact_amount + fact_tolerance_amount
-                              )
+                              ))
             # modify indices randomly
-            indices <- indices + runif( n = length(indices), 
-                              min = - fact_tolerance_amount,
-                              max = + fact_tolerance_amount
-                              )
+            indices <- indices + round(runif( n = length(indices), 
+                              min = - fact_tolerance_day,
+                              max = + fact_tolerance_day
+                              ))
         } else {
             if(worst_case) {
                 # worst case: all costs are highest
@@ -123,6 +158,7 @@ fact_amounts_series <- function(
                 # best case: all incomes are earliest
                 indices <- indices - sign(fact_amount) * fact_tolerance_day
             }
+        }
         # fix indices that lie outside the output vector
         # cat("indices before fixing: ",indices,"\n")
         # cat("amounts before fixing: ",amounts,"\n")
@@ -140,13 +176,16 @@ fact_amounts_series <- function(
         }
         # cat("indices after fixing: ",indices,"\n")
         # cat("amounts after fixing: ",amounts,"\n")
-        }
     } else {
         # keep amount
         amounts <- rep(fact_amount, length(indices))
     }
     # cat("amounts before putting into out: ",amounts,"\n")
 
+    # cat("indices: ",indices,"\n")
+    # cat("amounts: ",amounts,"\n")
+    stopifnot(all(1<=indices|indices<=N))
+    stopifnot(length(indices) == length(amounts))
     # set the amounts to the indices
     out[indices] <- amounts
     # cat("out: ",out,"\n")
@@ -155,39 +194,6 @@ fact_amounts_series <- function(
     return(out)
 }
 
-budget_ensemble<- function( budget,  
-    start = Sys.Date(), end = Sys.Date() + 365,
-    ensemble_size = 100
-    ) {
-    # run without tolerance
-    timeseries_without_tolerance <- timeseries_from_budget(
-        budget = budget, start = start, end = end, with_tolerance = TRUE, 
-        random_tolerance = FALSE)
-    # the ensemble out starts with the bare run
-    ENSEMBLE_OUT <- timeseries_without_tolerance
-    if(any(is.finite(budget$tolerance_amount))) {
-        # create ensemble matrix
-        ENSEMBLE <- matrix(NA,nrow=ensemble_size, ncol = nrow(ENSEMBLE_OUT))
-        # do the runs
-        # cat("create members...")
-        for(i in 1:ensemble_size) {
-            ENSEMBLE[i,] <- timeseries_from_budget( budget = budget, 
-                        start = start, end = end, 
-                        with_tolerance = TRUE, random_tolerance=TRUE)$amount
-        }
-        cat("done!\n")
-        # calculate statistics
-        # cat("calculate statistics...")
-        ENSEMBLE_OUT$ensmean <- apply(X = ENSEMBLE,MARGIN = 2, FUN = mean)
-        ENSEMBLE_OUT$ensmedian <- apply(X = ENSEMBLE,MARGIN = 2, FUN = median)
-        ENSEMBLE_OUT$ensquant25 <- apply(X = ENSEMBLE,MARGIN = 2, FUN = function(x)quantile(x,probs = c(0.25)))
-        ENSEMBLE_OUT$ensquant75 <- apply(X = ENSEMBLE,MARGIN = 2, FUN = function(x)quantile(x,probs = c(0.75)))
-        ENSEMBLE_OUT$ensmin <- apply(X = ENSEMBLE,MARGIN = 2, FUN = min)
-        ENSEMBLE_OUT$ensmax <- apply(X = ENSEMBLE,MARGIN = 2, FUN = max)
-        # cat("done!\n")
-    }
-    return(ENSEMBLE_OUT)
-}
 
 plot_budget_timeseries <- function(timeseries) {
     plotrange <- range(c(timeseries$amount,timeseries$worstcase,
@@ -223,7 +229,32 @@ plot_budget_timeseries <- function(timeseries) {
     if(!is.null(timeseries$worstcase) & !is.null(timeseries$bestcase)) {
         polygon(x = c(timeseries$day,rev(timeseries$day)), 
                 y = c(timeseries$worstcase,rev(timeseries$bestcase)),
-                col = "#00000022",border=NA)
+                col = "#00000033",border=NA)
+    }
+    if(!is.null(timeseries$ensmin) & !is.null(timeseries$ensmax)) {
+        polygon(x = c(timeseries$day,rev(timeseries$day)), 
+                y = c(timeseries$ensmin,rev(timeseries$ensmax)),
+                col = "#00000033",border=NA)
+    }
+    if(!is.null(timeseries$ensquant05) & !is.null(timeseries$ensquant95)) {
+        polygon(x = c(timeseries$day,rev(timeseries$day)), 
+                y = c(timeseries$ensquant05,rev(timeseries$ensquant95)),
+                col = "#00000033",border=NA)
+    }
+    if(!is.null(timeseries$ensquant10) & !is.null(timeseries$ensquant90)) {
+        polygon(x = c(timeseries$day,rev(timeseries$day)), 
+                y = c(timeseries$ensquant10,rev(timeseries$ensquant90)),
+                col = "#00000033",border=NA)
+    }
+    if(!is.null(timeseries$ensquant25) & !is.null(timeseries$ensquant75)) {
+        polygon(x = c(timeseries$day,rev(timeseries$day)), 
+                y = c(timeseries$ensquant25,rev(timeseries$ensquant75)),
+                col = "#00000033",border=NA)
+    }
+    if(!is.null(timeseries$ensmean)) {
+        lines(x = timeseries$day, y = timeseries$ensmean
+              ,lwd = 2, lty = 2
+              )
     }
     # raw run
     lines(x = timeseries$day, y = timeseries$amount
@@ -240,7 +271,7 @@ plot_budget_timeseries_to_png <- function(timeseries,filename,width=600,height=4
 
 #### read data ####
 # BUDGET <- read_budget_from_text(readLines("~/Downloads/budget.simbuto"))
-# MONEY <- timeseries_from_budget(budget = BUDGET)
+# MONEY <- timeseries_from_budget(budget = BUDGET, ensemble_size = 500)
 # cat("plotting...")
 # plot_budget_timeseries(MONEY)
 # cat("done!\n")