Goog's image
Posts 4
Joined 21 Sep '12
Email User

I find that

  • Member 36172845 spent 15 days in hospital in Y3 but had no claim in that year.
  • Member 35838713 had a claim in Y3 that has length 4-8 weeks in Impatient Hospital. However, that member doesn't exist in the Y3 days-in-hospital report.
Those members are just examples from the data set.
Do I mis-interpreret the data? Can someone explain?
 
Goog's image
Posts 4
Joined 21 Sep '12
Email User

Aren't you concerned by this?

 
DavidChudzicki's image
DavidChudzicki
Kaggle Admin
Posts 447
Thanks 107
Joined 21 Nov '10
Email User
From Kaggle

Incomplete data is a fact of life -- we try to make things nicer for you than if you were dealing with data in its original form, but some this kind of thing is really just part of the game.

Thanked by Sarkis
 
Sarkis's image
Posts 41
Thanks 5
Joined 5 Apr '11
Email User

Goog wrote:

I find that

  • Member 36172845 spent 15 days in hospital in Y3 but had no claim in that year.
  • Member 35838713 had a claim in Y3 that has length 4-8 weeks in Impatient Hospital. However, that member doesn't exist in the Y3 days-in-hospital report.
Those members are just examples from the data set.
Do I mis-interpreret the data? Can someone explain?

We need to predict days in hospital in year Y4 based on the claims data from prior years. That's why it's ok if a member has entry in DaysInHospital_Y3 but no claims data for Y3 or visa versa. See also: http://www.heritagehealthprize.com/c/hhp/forums/t/651/bad-data

Thanked by DavidChudzicki and Goog
 
Goog's image
Posts 4
Joined 21 Sep '12
Email User

DavidChudzicki wrote:

Incomplete data is a fact of life -- we try to make things nicer for you than if you were dealing with data in its original form, but some this kind of thing is really just part of the game.

It wasn't about incomplete data. What I was questioning is about data inconsistency. Days in hospital data doesn't match claims data.

It must mean that I somehow misinterpreted the meaning of days in hospital.

 
Goog's image
Posts 4
Joined 21 Sep '12
Email User

Sarkis wrote:

We need to predict days in hospital in year Y4 based on the claims data from prior years. That's why it's ok if a member has entry in DaysInHospital_Y3 but no claims data for Y3 or visa versa. See also: http://www.heritagehealthprize.com/c/hhp/forums/t/651/bad-data

I've read the conversation you provided. I think that the original poster had the same concern but the reply posts didn't answer his question. So I'd like to make clear it here.

Yes, our job is to predict DiH in Y4. However, what are we gonna do if we don't know what days in hospital means?

According to this explanation by HHP, DaysInHospital(Member, Year) = Sum(Claim.LengthOfStay) for Member's claims in Year where PlaceSVC is not "Urgent Care" or "Inpatient Hospital". However, that formula is NOT true for Y2 and Y3.

  • Member 36172845 spent 15 days in hospital in Y3 but had no claim in that year.
  • Member 35838713 had a claim in Y3 that has length 4-8 weeks in Impatient Hospital. However, that member doesn't exist in the Y3 days-in-hospital report.
 
David J. Slate's image
Rank 10th
Posts 85
Thanks 29
Joined 5 Aug '10
Email User

Sorry, somehow my post ended up on the wrong thread.

 

Reply

Flag alert Flagging is a way of notifying administrators that this message contents inappropriate or abusive content. Are you sure this forum post qualifies?