+
+
+
+
+Data Cleaning Exercise
+1. Outlier in MVPA
+Somebody has a daily MVPA value of 1352.215, which equates to roughly 22.5 hours of moderate to vigorous physical activity per day, over the past 7 days. This value is highly suspect.
+ 1
+ 2
+ 3
+ 4
+ 5
+ 6
+ 7
+ 8
+ 9
+10
+11
+12
+13
+14
+15
+16 | . sum daily_mvpa, d
+
+ daily_mvpa
+-------------------------------------------------------------
+ Percentiles Smallest
+ 1% 2.083333 .6666667
+ 5% 5.8 1
+10% 7.9 1.5 Obs 526
+25% 13.3 1.5 Sum of Wgt. 526
+
+50% 22.96429 Mean 29.78098
+ Largest Std. Dev. 60.75801
+75% 37.33333 99.85714
+90% 53.58333 107.0714 Variance 3691.536
+95% 64.25 133.7 Skewness 19.68991
+99% 92.16666 1352.215 Kurtosis 428.3505
+
|
+2. Calorie Consumption
+There are some individuals that have very low values of calorie consumption. It is highly unlikely that anybody could routinely consume fewer than 500 calories per day. Upon further examination, this includes individuals who are classified as overweight (BMI > 25) and obese (BMI > 30).
+
+3. Missingness
+This data set contains many missing values. Of the 615 observations, 473 are missing self-reported physical activity, 3 are missing BMI, 202 are missing calorie information, 90 are missing accelerometer data, and 522 are missing Arizona physical activity data.
+4. Variable Concordance
+There is little concordance between the Arizona PA values and the actual accelerometer values. The correlation is quite low (r=0.13) and we can see some cases where the MET values are quite high for the actual amount of MVPA achieved. There does not need to be perfect concordance between these measures as they capture different aspects of PA, but the very low correlation raises suspicion.
+
+The concordance between self-reported PA and accelerometer PA is even more disconcerting. It looks like there is almost no relationship between the two.
+5. Macronutrients
+Seemingly the percent of dietary intake due to macronutrients can be computed as 100% = (percent fat) + (percent protein) + (percent carbohydrates). However, when computing the percent of each macronutrient as (macronutrient grams)(multiplier)/(total calories), the values do not add up to 100%.
+One individual seems to get 100% of their calories from fat, which is highly unlikely.
+
+
+
+
+
+
+
+
+
+
+
+
+
+