Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Calculating difference between observations in a panel with missing values

    Hello!

    I am working with online auction data (see -dataex- below), where id is the listing identifier, id_user is the auction participant identifier, datetime is time stamp, bid = 1 if observation is a bid vs. comment (=0), and bid_it is a bid (in $). The auction runs for 7 days and at any time any user can either make a bid or leave a comment. That is why there are missing values between the bids

    For each of the bids, what would be the appropriate way to calculate the difference between the current and previous bids (vs. absolute monetary values)?

    Thanks!

    Code:
    * Example generated by -dataex-. For more info, type help dataex
    clear
    input int id float id_user double datetime float bid long bid_it
    2 12613 1952878380000 1 66500
    2     . 1952878440000 1     .
    2 12613 1952878440000 1 67500
    2  9609 1.9528785e+12 1 68000
    2 24717 1952878620000 0     .
    2 12613 1952878620000 1 69500
    2  3352 1952878620000 1 69000
    2 11882 1952878680000 0     .
    2  5980 1952878740000 0     .
    2 18853 1952878740000 0     .
    2  3364 1952878740000 1 70000
    2 12613 1.9528788e+12 1 70500
    2 12613 1952878860000 0     .
    2  9099 1952878860000 0     .
    2  3352 1952878920000 1 71000
    2  3364 1952878980000 1 71500
    2 15927 1952879040000 1 72211
    2  3352 1.9528791e+12 1 75211
    2  9609 1952879160000 0     .
    2 11882 1952879220000 0     .
    2 12613 1952879220000 1 76211
    2 11882 1952879280000 0     .
    2 18853 1952879280000 0     .
    2 12613 1952879280000 1 78211
    2  3364 1952879280000 1 77211
    2  9609 1952879340000 0     .
    2  5980 1952879340000 0     .
    2  3364 1.9528794e+12 1 79211
    2 12613 1952879460000 1 79711
    2 21154 1952879580000 1 80711
    2  3352 1952879580000 1 80211
    2 12565 1.9528797e+12 0     .
    2     . 1.9528797e+12 1     .
    2 11267 1952879760000 0     .
    2 24717 1952879760000 0     .
    2  3352 1952879760000 0     .
    2  9099 1952879820000 0     .
    2 11267 1952879880000 0     .
    2 11267 1952880120000 0     .
    2 12613 1952881080000 0     .
    3 17377 1952283540000 0     .
    3   914 1952283540000 0     .
    3 11176 1952283780000 0     .
    3 24784 1952284440000 0     .
    3   345 1952284980000 0     .
    3 15933 1952285160000 0     .
    3  4324  1.952286e+12 0     .
    3  4324  1.952286e+12 1 20000
    3  4541 1952286120000 0     .
    3 13497 1952291040000 0     .
    3  4541 1952292420000 0     .
    3 14562 1952311380000 1 20250
    3 14562 1952311440000 0     .
    3 23221 1952318940000 0     .
    3  6403 1952322420000 0     .
    3 17926 1952323680000 0     .
    3 17896 1952326620000 0     .
    3  4262 1952327640000 0     .
    3  4541 1952328660000 0     .
    3  1536 1952328780000 0     .
    3   139 1952329740000 0     .
    3 24707 1952329920000 0     .
    3  4541 1952329980000 0     .
    3 23436 1952330280000 0     .
    3  9997 1952330280000 0     .
    3   914 1.9523304e+12 0     .
    3  9086 1952331180000 0     .
    3 23436 1952331360000 0     .
    3 23424 1952331960000 0     .
    3 23436 1952332140000 0     .
    3 26316 1952336640000 0     .
    3 16396 1.9523373e+12 0     .
    3  4541 1952339340000 0     .
    3 23701 1952343180000 1 21000
    3  8933 1.9523445e+12 0     .
    3  3613 1952353020000 0     .
    3  4285 1952353920000 0     .
    3  2638 1952354340000 0     .
    3  5746 1952358060000 0     .
    3  4541 1952358780000 0     .
    3 23436 1952363220000 0     .
    3  7012 1952368920000 0     .
    3 22082 1952372040000 0     .
    3 21217 1952374320000 0     .
    3  1536 1952377020000 0     .
    3  1622 1952380020000 0     .
    3 11950 1952412060000 0     .
    3 23436 1.9524147e+12 0     .
    3  4541 1952422920000 0     .
    3 17165 1952433180000 0     .
    3 14629 1952442540000 1 22000
    3  4123 1952446440000 1 23426
    3 15638 1952446740000 0     .
    3  7329 1952502420000 0     .
    3  9907 1952513280000 0     .
    3  9907 1952513340000 1 24426
    3  6603 1952534520000 0     .
    3  4541 1952544060000 0     .
    3   853 1952548080000 1 25555
    3 23518   1.95255e+12 0     .
    end
    format %tcNN/DD/CCYY_HH:MM datetime
    Last edited by Anton Ivanov; 21 Nov 2021, 18:07.

  • #2
    I'm not sure I fully understand your explanation and data, but perhaps what you want is this:
    Code:
    by id (datetime), sort: gen last_bid = 0 if _n == 1
    by id (datetime): replace last_bid = cond(!missing(bid_it[_n-1]), bid_it[_n-1], ///
        last_bid[_n-1])
    gen increment = bid_it - last_bid
    My concerns are:

    1. it would seem that if bid == 1, there should always be a value of bid_it, but there are exceptions to that in observatoins 2 and 33.

    2. I imagine that the rules of an auction require that each bid exceed the preceding one, but, again, that is not always the case here.

    If those do not represent errors in your data, which you should fix, can you clarify for me why my expectations here are off base? Thanks.

    Comment


    • #3
      Clyde Schechter Thank you for your response, Clyde. The suggested code worked perfectly.

      1. Thank you for pointing out the missing values in bid_it for bid = 1. I am aware of this issue, which comes as a result of improper import of very long strings. I am working on it.
      2. In case of this particular auction, the minimum bid is $100.

      If that still matters, can you please clarify what you mean under "my expectations here are off base". Thanks!

      Comment


      • #4
        Well, you clarified #1 for me. But, not that I have ever participated in one, but my understanding of auctions is that each bidder must bid more than the previous bid. But there are some exceptions to that in your data. If those are not data errors, I'm wondering what the rules actually are.

        Comment


        • #5
          Clyde Schechter Now I see where you are pointing -- some of the higher values appearing before the lower ones. Included below is the corrected version with increment calculation following your approach.

          Another question I have: Is there any feasible way to organize such data in the longitudinal panel format? For example, given that the action runs for 7 days (could start at any of the days) I first created the day of the week (dow) time stamp. And then aggregated data of interest by id_user and dow (then -xtset-). Limitations of such approach include but are not limited to missing values for days when there are no comments/bids, or bias in averaging two or more bids for a given day. Thanks for feedback.

          Code:
          * Example generated by -dataex-. For more info, type help dataex
          clear
          input int id double datetime float(id_user bid) long bid_it float increment
          2 1952878740000  5980 0     .    .
          2 1.9528788e+12 12613 1 70500  500
          2 1952878860000 12613 0     .    .
          2 1952878860000  9099 0     .    .
          2 1952878920000  3352 1 71000  500
          2 1952878980000  3364 1 71500  500
          2 1952879040000 15927 1 72211  711
          2 1.9528791e+12  3352 1 75211 3000
          2 1952879160000  9609 0     .    .
          2 1952879220000 12613 1 76211 1000
          2 1952879220000 11882 0     .    .
          2 1952879280000  3364 1 77211 1000
          2 1952879280000 12613 1 78211 1000
          2 1952879280000 11882 0     .    .
          2 1952879280000 18853 0     .    .
          2 1952879340000  9609 0     .    .
          2 1952879340000  5980 0     .    .
          2 1.9528794e+12  3364 1 79211 1000
          2 1952879460000 12613 1 79711  500
          2 1952879580000  3352 1 80211  500
          2 1952879580000 21154 1 80711  500
          2 1.9528797e+12 12565 0     .    .
          2 1952879760000 11267 0     .    .
          2 1952879760000 24717 0     .    .
          2 1952879760000  3352 0     .    .
          2 1952879820000  9099 0     .    .
          2 1952879880000 11267 0     .    .
          2 1952880120000 11267 0     .    .
          2 1952881080000 12613 0     .    .
          3 1952283540000   914 0     .    .
          3 1952283540000 17377 0     .    .
          3 1952283780000 11176 0     .    .
          3 1952284440000 24784 0     .    .
          3 1952284980000   345 0     .    .
          3 1952285160000 15933 0     .    .
          3  1.952286e+12  4324 1 20000    .
          3  1.952286e+12  4324 0     .    .
          3 1952286120000  4541 0     .    .
          3 1952291040000 13497 0     .    .
          3 1952292420000  4541 0     .    .
          3 1952311380000 14562 1 20250  250
          3 1952311440000 14562 0     .    .
          3 1952318940000 23221 0     .    .
          3 1952322420000  6403 0     .    .
          3 1952323680000 17926 0     .    .
          3 1952326620000 17896 0     .    .
          3 1952327640000  4262 0     .    .
          3 1952328660000  4541 0     .    .
          3 1952328780000  1536 0     .    .
          3 1952329740000   139 0     .    .
          3 1952329920000 24707 0     .    .
          3 1952329980000  4541 0     .    .
          3 1952330280000  9997 0     .    .
          3 1952330280000 23436 0     .    .
          3 1.9523304e+12   914 0     .    .
          3 1952331180000  9086 0     .    .
          3 1952331360000 23436 0     .    .
          3 1952331960000 23424 0     .    .
          3 1952332140000 23436 0     .    .
          3 1952336640000 26316 0     .    .
          3 1.9523373e+12 16396 0     .    .
          3 1952339340000  4541 0     .    .
          3 1952343180000 23701 1 21000  750
          3 1.9523445e+12  8933 0     .    .
          3 1952353020000  3613 0     .    .
          3 1952353920000  4285 0     .    .
          3 1952354340000  2638 0     .    .
          3 1952358060000  5746 0     .    .
          3 1952358780000  4541 0     .    .
          3 1952363220000 23436 0     .    .
          3 1952368920000  7012 0     .    .
          3 1952372040000 22082 0     .    .
          3 1952374320000 21217 0     .    .
          3 1952377020000  1536 0     .    .
          3 1952380020000  1622 0     .    .
          3 1952412060000 11950 0     .    .
          3 1.9524147e+12 23436 0     .    .
          3 1952422920000  4541 0     .    .
          3 1952433180000 17165 0     .    .
          3 1952442540000 14629 1 22000 1000
          3 1952446440000  4123 1 23426 1426
          3 1952446740000 15638 0     .    .
          3 1952502420000  7329 0     .    .
          3 1952513280000  9907 0     .    .
          3 1952513340000  9907 1 24426 1000
          3 1952534520000  6603 0     .    .
          3 1952544060000  4541 0     .    .
          3 1952548080000   853 1 25555 1129
          3   1.95255e+12 23518 0     .    .
          3 1952590920000  4541 0     .    .
          3 1952601060000 21842 1 26426  871
          3 1952619720000 14536 0     .    .
          3 1952622720000  4541 0     .    .
          3 1.9527558e+12  2811 0     .    .
          3 1952762640000  4541 0     .    .
          3 1952766120000 23191 1 27000  574
          3 1952769840000  4541 0     .    .
          3 1952807580000  8740 1 27750  750
          3 1952843340000  8534 0     .    .
          3 1.9528455e+12 12692 1 35000 7250
          end
          format %tcNN/DD/CCYY_HH:MM datetime

          Comment


          • #6
            I don't see this data as having true panel structure. It is far more complicated than that. You have items up for auction (id) and you have bidders (id_user) in a complex relationship: they are neither nested nor crossed. This is called a multiple membership structure. Moreover for any id, the same id_user can submit multiple bids over time. So we have repeated observations nested in this multiple membership structure. Any attempt to crush this into a panel structure is bound to discard information (for example, when you average bids by the same bidder in the same day). Now, there may be particular questions that can be posed and answered with a simplified structure and does not need all of the information. But unless you have something specific of that nature in mind, I think the data organization you have is probably optimal.

            Comment


            • #7
              Clyde Schechter Thank you for your feedback, Clyde. As a matter of fact, there is something specific in my mind in relation to the structure of the data. My theory tells me (and this is something I would like to test) that the bid at time t is impacted by the comments preceding it. For example, say, there is a new listing and we observe the following dynamics between comments and bids:

              id_user1: ok, i am first. what a gorgeous little car
              id_user2: i’m already jealous of the new owner, whoever it will be.
              id_user3: $6,900 bid placed
              id_user4: What a beautiful example of this fantastic (and often underrated) car!
              id_user5: $8,500 bid placed
              id_user6: Can you please post a video of the top going up and down? Speaking of the top, have the opening mechanism been serviced (i.e. rams checked, alignment etc.)
              id_user7: Also, what services were performed on the car since ownership? Service A B C?
              id_user8: $9,000 bid placed
              id_user9: $9,500 bid placed
              id_user10: CarFax is noting vehicle color was Silver ?? Please explain.

              For simplicity, let's assume that sentiment of the comments is my measurement of interest and it is expected to positively impact the bid (or increment). Then, I would expect:

              t: Mean_sentiment(id_user1+id_user2) --> Bid(id_user3)
              t+1: Sentiment(id_user4) --> Bid(id_user5)
              t+2: Mean_sentiment(id_user6+id_user7) --> Bid(id_user8) /// Alternatively, I could consider Mean_bid(id_user8+id_user9)

              Does such approach seem feasible?
              Last edited by Anton Ivanov; 22 Nov 2021, 13:07.

              Comment


              • #8
                This is way out of my field. Speaking only as an educated layperson, this looks very challenging. You have previous comments influencing bids, which may in turn influence subsequent comments, which may, in turn, influence subsequent bids, and on and on. Moreover, a bid at a given time may also be influenced by comments made much earlier in the course of the auction. Then there is the question of how you extract the sentiment of a comment into a numeric variable that can be used for analysis--maybe there's some validated standard way to do this, but I'm not familiar with any. Seeing what you have in mind, I think that if this is going to be possible at all, you will need to squeeze the most you can from your data, so that trying to compress it into a conventional panel format strikes me as a potentially fatal mistake.

                Again, this is not my domain of expertise at all, and maybe there are standard approaches that overcome all of these challenges and I'm just not aware of them. But from an outsider's perspective this looks, to put it positively, very ambitious. To put it less positively, daunting and perilous.

                Comment


                • #9
                  Clyde Schechter Thank you for your feedback, Clyde. I agree with you that simultaneity is a significant concern given the dynamic relationship between comments and bids.

                  To better understand the underlying mechanism I would still need some form of the panel dataset. To this end, two possible approaches come to mind: one assuming each individual bid as point in time, and another assuming each group of bids as point in time. Such structure assumes the mean cumulative effect of the preceding quality of content captured along with bids at time t.

                  To elaborate on the possible data structure behind either of the approaches I am going to use the exemplar data from #5 above plus one more additional variable wc_it, which represents content quality measured in terms of word count. To facilitate comprehension of my elaborations, please refer to the attached screenshot and actual Excel work sheet. The latter approach highlighted in green, which treats group of bids (their mean) as point in time, seems to be somewhat optimal in terms of preserving both information and longitudinal structure. Do you think such solution is plausible and Stata can handle the data preparation?

                  Click image for larger version

Name:	image_24847.png
Views:	1
Size:	620.9 KB
ID:	1637853
                  Attached Files
                  Last edited by Anton Ivanov; 23 Nov 2021, 15:39.

                  Comment


                  • #10
                    Your screen shot is not readable on my computer. And I do not download attachments from people I don't know--especially not Office documents, which can contain active content. As you clearly know, the effective way to share example data here is with -dataex-.

                    Comment


                    • #11
                      Clyde Schechter My apologies for the inconvenience. Below is the resulting -dataex-:

                      Code:
                      * Example generated by -dataex-. For more info, type help dataex
                      clear
                      input byte id double datetime int id_user byte bid long bid_it int wc_it byte time1 long H double(wc_it_mean_zeros wc_it_mean_cum) byte time2 double(bid_it_mean wc_it_mean)
                      2      1952878740000  5980 0     .   5  .     .                  .                  .  .       .                  .
                      2 1952878800000.0002 12613 1 70500   .  1 70500                  5                  5  1   70500                  5
                      2 1952878859999.9998 12613 0     .  19  .     .                  .                  .  .       .                  .
                      2 1952878859999.9998  9099 0     .  17  .     .                  .                  .  .       .                  .
                      2      1952878920000  3352 1 71000   .  2 71000                 18                 18  2 72480.5                 18
                      2      1952878980000  3364 1 71500   .  3 71500                  0                 18  .       .                  .
                      2 1952879040000.0002 15927 1 72211   .  4 72211                  0                 18  .       .                  .
                      2 1952879099999.9998  3352 1 75211   .  5 75211                  0                 18  .       .                  .
                      2      1952879160000  9609 0     .  19  .     .                  .                  .  .       .                  .
                      2 1952879220000.0002 12613 1 76211   .  6 76211                 19                 19  3   76211                 19
                      2 1952879220000.0002 11882 0     .  14  .     .                  .                  .  .       .                  .
                      2 1952879279999.9998  3364 1 77211   .  7 77211                 14                 14  4   77711                 14
                      2 1952879279999.9998 12613 1 78211   .  8 78211                  0                 14  .       .                  .
                      2 1952879279999.9998 11882 0     .   5  .     .                  .                  .  .       .                  .
                      2 1952879279999.9998 18853 0     .   7  .     .                  .                  .  .       .                  .
                      2      1952879340000  9609 0     .  21  .     .                  .                  .  .       .                  .
                      2      1952879340000  5980 0     .   7  .     .                  .                  .  .       .                  .
                      2 1952879400000.0002  3364 1 79211   .  9 79211                 10                 10  5   79961                 10
                      2 1952879459999.9998 12613 1 79711   . 10 79711                  0                 10  .       .                  .
                      2 1952879580000.0002  3352 1 80211   . 11 80211                  0                 10  .       .                  .
                      2 1952879580000.0002 21154 1 80711   . 12 80711                  0                 10  .       .                  .
                      2      1.9528797e+12 12565 0     .   1  .     .                  .                  .  .       .                  .
                      2 1952879760000.0002 11267 0     .  24  .     .                  .                  .  .       .                  .
                      2 1952879760000.0002 24717 0     .   1  .     .                  .                  .  .       .                  .
                      2 1952879760000.0002  3352 0     .   4  .     .                  .                  .  .       .                  .
                      2 1952879819999.9998  9099 0     .  16  .     .                  .                  .  .       .                  .
                      2      1952879880000 11267 0     .  24  .     .                  .                  .  .       .                  .
                      2      1952880120000 11267 0     .   6  .     .                  .                  .  .       .                  .
                      2      1952881080000 12613 0     .  27  .     .                  .                  .  .       .                  .
                      3 1952283539999.9998   914 0     .   9  .     .                  .                  .  .       .                  .
                      3 1952283539999.9998 17377 0     .   5  .     .                  .                  .  .       .                  .
                      3      1952283780000 11176 0     .   3  .     .                  .                  .  .       .                  .
                      3 1952284440000.0002 24784 0     .   6  .     .                  .                  .  .       .                  .
                      3      1952284980000   345 0     .  15  .     .                  .                  .  .       .                  .
                      3      1952285160000 15933 0     .  12  .     .                  .                  .  .       .                  .
                      3 1952286000000.0002  4324 1 20000   .  1 20000  8.333333333333334  8.333333333333334  1   20000  8.333333333333334
                      3 1952286000000.0002  4324 0     .  26  .     .                  .                  .  .       .                  .
                      3      1952286120000  4541 0     .  25  .     .                  .                  .  .       .                  .
                      3 1952291039999.9998 13497 0     .  44  .     .                  .                  .  .       .                  .
                      3 1952292419999.9998  4541 0     .  49  .     .                  .                  .  .       .                  .
                      3      1952311380000 14562 1 20250   .  2 20250                 36                 36  2   20250                 36
                      3 1952311440000.0002 14562 0     .   9  .     .                  .                  .  .       .                  .
                      3 1952318940000.0002 23221 0     .  34  .     .                  .                  .  .       .                  .
                      3 1952322420000.0002  6403 0     .  33  .     .                  .                  .  .       .                  .
                      3      1952323680000 17926 0     .   7  .     .                  .                  .  .       .                  .
                      3      1952326620000 17896 0     .   7  .     .                  .                  .  .       .                  .
                      3 1952327640000.0002  4262 0     .  30  .     .                  .                  .  .       .                  .
                      3 1952328659999.9998  4541 0     .  30  .     .                  .                  .  .       .                  .
                      3 1952328780000.0002  1536 0     .  13  .     .                  .                  .  .       .                  .
                      3 1952329740000.0002   139 0     .   9  .     .                  .                  .  .       .                  .
                      3      1952329920000 24707 0     .  44  .     .                  .                  .  .       .                  .
                      3 1952329980000.0002  4541 0     .  38  .     .                  .                  .  .       .                  .
                      3      1952330280000  9997 0     .  64  .     .                  .                  .  .       .                  .
                      3      1952330280000 23436 0     .  44  .     .                  .                  .  .       .                  .
                      3 1952330399999.9998   914 0     .  17  .     .                  .                  .  .       .                  .
                      3 1952331179999.9998  9086 0     .   8  .     .                  .                  .  .       .                  .
                      3 1952331359999.9998 23436 0     .  82  .     .                  .                  .  .       .                  .
                      3 1952331959999.9998 23424 0     .  13  .     .                  .                  .  .       .                  .
                      3 1952332139999.9998 23436 0     .  51  .     .                  .                  .  .       .                  .
                      3      1952336640000 26316 0     .  27  .     .                  .                  .  .       .                  .
                      3 1952337300000.0002 16396 0     .  10  .     .                  .                  .  .       .                  .
                      3      1952339340000  4541 0     .  13  .     .                  .                  .  .       .                  .
                      3      1952343180000 23701 1 21000   .  3 21000 27.761904761904763 27.761904761904763  3   21000 27.761904761904763
                      3 1952344499999.9998  8933 0     .  46  .     .                  .                  .  .       .                  .
                      3      1952353020000  3613 0     . 105  .     .                  .                  .  .       .                  .
                      3 1952353919999.9998  4285 0     .  10  .     .                  .                  .  .       .                  .
                      3 1952354339999.9998  2638 0     .  13  .     .                  .                  .  .       .                  .
                      3      1952358060000  5746 0     .  19  .     .                  .                  .  .       .                  .
                      3      1952358780000  4541 0     .  43  .     .                  .                  .  .       .                  .
                      3      1952363220000 23436 0     .   2  .     .                  .                  .  .       .                  .
                      3 1952368920000.0002  7012 0     .   5  .     .                  .                  .  .       .                  .
                      3 1952372039999.9998 22082 0     .   7  .     .                  .                  .  .       .                  .
                      3 1952374320000.0002 21217 0     .   5  .     .                  .                  .  .       .                  .
                      3 1952377020000.0002  1536 0     .  35  .     .                  .                  .  .       .                  .
                      3      1952380020000  1622 0     .  12  .     .                  .                  .  .       .                  .
                      3      1952412060000 11950 0     .   6  .     .                  .                  .  .       .                  .
                      3 1952414699999.9998 23436 0     .  56  .     .                  .                  .  .       .                  .
                      3 1952422920000.0002  4541 0     .  38  .     .                  .                  .  .       .                  .
                      3 1952433179999.9998 17165 0     .  15  .     .                  .                  .  .       .                  .
                      3      1952442540000 14629 1 22000   .  4 22000            26.0625            26.0625  4   22713            26.0625
                      3 1952446440000.0002  4123 1 23426   .  5 23426                  0            26.0625  .       .                  .
                      3      1952446740000 15638 0     .  15  .     .                  .                  .  .       .                  .
                      3 1952502419999.9998  7329 0     .   7  .     .                  .                  .  .       .                  .
                      3      1952513280000  9907 0     .  20  .     .                  .                  .  .       .                  .
                      3 1952513340000.0002  9907 1 24426   .  6 24426                 14                 14  5   24426                 14
                      3      1952534520000  6603 0     .  44  .     .                  .                  .  .       .                  .
                      3 1952544059999.9998  4541 0     .  27  .     .                  .                  .  .       .                  .
                      3 1952548080000.0002   853 1 25555   .  7 25555               35.5               35.5  6   25555               35.5
                      3 1952550000000.0002 23518 0     .  19  .     .                  .                  .  .       .                  .
                      3 1952590919999.9998  4541 0     .  39  .     .                  .                  .  .       .                  .
                      3      1952601060000 21842 1 26426   .  8 26426                 29                 29  7   26426                 29
                      3      1952619720000 14536 0     .  26  .     .                  .                  .  .       .                  .
                      3 1952622720000.0002  4541 0     . 105  .     .                  .                  .  .       .                  .
                      3 1952755800000.0002  2811 0     .  42  .     .                  .                  .  .       .                  .
                      3      1952762640000  4541 0     .  33  .     .                  .                  .  .       .                  .
                      3      1952766120000 23191 1 27000   .  9 27000               51.5               51.5  8   27000               51.5
                      3 1952769840000.0002  4541 0     .  22  .     .                  .                  .  .       .                  .
                      3      1952807580000  8740 1 27750   . 10 27750                 22                 22  9   27750                 22
                      3 1952843340000.0002  8534 0     .  13  .     .                  .                  .  .       .                  .
                      3 1952845499999.9998 12692 1 35000   . 11 35000                 13                 13 10   35000                 13
                      end
                      format %tcnn/dd/ccYY_hh:MM datetime
                      Last edited by Anton Ivanov; 24 Nov 2021, 08:49.

                      Comment


                      • #12
                        I'm not sure. Looking at the data, it isn't apparent to me how you arrived at your grouping of bids. The groups are not consistent in terms of either number of bids included, nor time interval they span. If there is some method to it that will persuade people that bids grouped together have something in common that they do not share with other bids, and that this grouping is sensible in the context of studying the influence of comments on bids, then it could be made to work computationally. But I don't grasp the underlying logic of it. Then again, this is not even close to my area of work, and it would be quite easy for me to miss the obvious here. I do think an opinion from somebody in your field would be more valuable than mine.

                        Comment

                        Working...
                        X