Grouping data by Month and Year? Example? about chartkick

[ankane]

I'd love to help, but I'm not sure I understand what you're trying to do. group_by_month should give you a time like 2012-01-01 00:00:00 UTC, and from that, you should be able to parse out the month and the year with Time.parse. Does that help at all?

[StephenOTT]

Looking to produce something like:

The other problem is about the ordering of dates within when dates are Hash Keys, and Adding missing dates. I have meeting looking for a gem but have had no luck.

For missing dates i am talking about imagine you had Jan, Feb, April, May, as data. But when you graph it you want to have march in as a value of 0.

I know i can handle missing dates by hand and sorting by hand, which I have been doing up to this point, but was just thinking there may be a gem already created for this type of manipulation.

[ankane]

The groupdate gem has the ability to fill in 0's if that's what you're using to generate the chart data https://github.com/ankane/groupdate#show-me-the-series-moneybag

[StephenOTT]

I am using MongoDB at the moment. according to ankane/groupdate#5 , mongo is not supported atm.

[StephenOTT]

@ankane Can you pass data manually into groupdate?

mmmm @ankane my problem seems to be because i am using MongoDB.

I am using the Aggregate Framework in Mongo to provide me with the counts of time series data such as the number of occurrences in a month:

    def analyzeRestaurantCreationDateCount
         restaurantCreationDateCount = @coll.aggregate([
            { "$match" => {fs_fcr_date: {"$ne" => nil}}},
            { "$project" => {createdate_month: {"$month" => "$fs_fcr_date"},createdate_year: {"$year" => "$fs_fcr_date"}}},
            { "$group" => {_id: {year:"$createdate_year", month:"$createdate_month"},number: { "$sum" => 1 }}},
            # { "$sort" => {"_id.year" => 1}},
        ])

        dateFormatChangeHash={}
        restaurantCreationDateCount.each do |x|
            temp = Date.strptime(Date::MONTHNAMES[x["_id"]["month"]] + " " + x["_id"]["year"].to_s, '%B %Y')
            dateFormatChangeHash[temp] = x["number"]    
        end

So when i get my data from mongo it already looks like:

{Mon, 01 Oct 2012=>53, Sat, 01 Dec 2012=>58, Mon, 01 Aug 2011=>8, Thu, 01 Nov 2012=>75, Thu, 01 Sep 2011=>58, Sun, 01 Sep 2013=>67, Sat, 01 Jun 2013=>561, Sat, 01 Sep 2012=>102, Wed, 01 May 2013=>609, Sun, 01 Jul 2012=>84, Wed, 01 Dec 2010=>1, Mon, 01 Jul 2013=>595, Fri, 01 Mar 2013=>253, Fri, 01 Jun 2012=>129, Wed, 01 Feb 2012=>95, Sun, 01 Apr 2012=>50, Thu, 01 Aug 2013=>736, Tue, 01 Jan 2013=>242, Tue, 01 Nov 2011=>46, Thu, 01 Mar 2012=>68, Sat, 01 Oct 2011=>48, Sun, 01 Jan 2012=>70, Wed, 01 Aug 2012=>70, Fri, 01 Feb 2013=>266, Thu, 01 Dec 2011=>47, Mon, 01 Apr 2013=>449, Tue, 01 May 2012=>77}

[ankane]

You won't be able to use groupdate for this.

However, I don't see where your code is failing. What happens when you pass the data you get from mongo into the line_chart helper?

[StephenOTT]

@ankane my data is presented correctly. BUT I want to show months that do not have values with a value of 0. So my understanding is that i need to add those months into the array.

I wrote up a loop to accomplish this:

Sample input data:
{Sun, 01 Jan 2012=>58, Sat, 01 Sep 2012=>53, Sat, 01 Dec 2012=>58}

and code snippet:

    def addMissingMonths (datesHash)
        count = 0

        datesHash.keys.each do |x|
            result = []
            if x != datesHash.keys.last
                (x+1.month).upto(datesHash.keys[count+1]-1.month) do |a|
                    result << [a.at_beginning_of_month]
                end
                puts result.uniq.to_s
            end
            count += 1
        end
    end

Thoughts?

Output at the moment for testing purposes looks like:

[[Wed, 01 Feb 2012], [Thu, 01 Mar 2012], [Sun, 01 Apr 2012], [Tue, 01 May 2012], [Fri, 01 Jun 2012], [Sun, 01 Jul 2012], [Wed, 01 Aug 2012]]
[[Mon, 01 Oct 2012], [Thu, 01 Nov 2012]]

And while going into prod the dates would be combined into the original hash with values of 0 and resorted

Thoughts?

https://gist.github.com/StephenOTT/6646927

[ankane]

Here's how groupdate does it: https://github.com/ankane/groupdate/blob/master/lib/groupdate/series.rb#L68

Essentially, it sorts the dates to find the earliest and latest. Then starting from the earliest, it adds months to an array until it hits the latest date.