Pure Pure Mint

[purepuremint]

Assignment: Making Data Management Decisions

1. My program

import pandas as pd import numpy as nu

# bug fix for display formats to avoid run time errors - put after code for loading data above pd.set_option(‘display.float_format’, lambda x:’%f’%x)

data = pd.read_csv('nesarc_pds.csv', low_memory=False)

#upper-case all DataFrame column names data.columns = map(str.upper, data.columns)

#setting variables you will be working with to numeric

data['S1Q2D'] = data['S1Q2D'].convert_objects(convert_numeric=True) data['S1Q2E'] = data['S1Q2E'].convert_objects(convert_numeric=True) data['S1Q2L'] = data['S1Q2L'].convert_objects(convert_numeric=True) data['S1Q2K'] = data['S1Q2K'].convert_objects(convert_numeric=True) data['S6Q1'] = data['S6Q1'].convert_objects(convert_numeric=True)

#subset data subdiv=data[(data['S1Q2E']>=3) & (data['S1Q2E']<=12) & (data['S1Q2D']==1)] subdie=data[(data['S1Q2L']>=3) & (data['S1Q2L']<=12) & (data['S1Q2K']==1)] #make a copy of my new subsetted data subdiv2 = subdiv.copy() subdie2 = subdie.copy() data2 = data.copy()

print ('counts for original S6Q1 (panic attack)') print(data2['S1Q2K'].value_counts(sort=False, dropna=False))

# recode missing values to python missing (NaN) data2['S6Q1'] = data2['S6Q1'].replace(9, nu.nan)

#if you want to include a count of missing add ,dropna=False after sort=False print ('counts for S6Q1 (panic attack) with 9 set to NAN and number of missing requested') print(data2['S6Q1'].value_counts(sort=False, dropna=False))

# categorize quantitative variable based on customized splits using cut function # splits into 3 groups (3-5, 6-9, 10-12) subdiv2['AGEDIV'] = pd.cut(subdiv2.S1Q2E, [2, 5, 9, 12]) subdie2['AGEDIE'] = pd.cut(subdie2.S1Q2L, [2, 5, 9, 12]) print(subdiv2['AGEDIV'].value_counts(sort=False, dropna=True)) print(subdie2['AGEDIE'].value_counts(sort=False, dropna=True))

#crosstabs evaluating which ages were put into which AGEGROUP3 print (pd.crosstab(subdiv2['AGEDIV'], subdiv2['S1Q2E'])) print (pd.crosstab(subdie2['AGEDIE'], subdie2['S1Q2L']))

#frequency distribution for AGEGROUP3 print ('counts for AGEDIV and AGEDIE') print(subdiv2['AGEDIV'].value_counts(sort=False)) print (subdie2['AGEDIE'].value_counts(sort=False))

print ('percentages for AGEDIV and AGEDIE') print(subdiv2['AGEDIV'].value_counts(sort=False, normalize=True)) print (subdie2['AGEDIE'].value_counts(sort=False, normalize=True))

2. The output that displays three of my variables as frequency tables + description comments

Counts for original S6Q1 (panic attack)

NaN     1036

1      4515

2     37358

9       184               (184 undefined responds)

Counts for S6Q1 (panic attack) with 9 set to NAN and number of missing requested NaN     1386          (undefined responds www replaced by NANs) 1      5278 2     36429 Cross tabs evaluating which ages were put into which age group S1Q2E     3    4    5    6    7    8    9    10   11   12 AGEDIV                                                   [2, 5]   420  374  562    0    0    0    0    0    0    0 [5, 9]     0    0    0  406  388  422  335    0    0    0 [9, 12]    0    0    0    0    0    0    0  478  332  538

S1Q2L     3    4    5    6    7    8    9    10   11   12 AGEDIE                                                   [2, 5]   218  177  225    0    0    0    0    0    0    0 [5, 9]     0    0    0  206  243  218  249    0    0    0 [9, 12]    0    0    0    0    0    0    0  251  227  337

Counts for AGEDIV and AGEDIE [2, 5]     1356 [5, 9]     1551 [9, 12]    1348 dtype: int64 [2, 5]     620 [5, 9]     916 [9, 12]    815 dtype: int64 Percentages for AGEDIV and AGEDIE [2, 5]     0.318684 [5, 9]     0.364512 [9, 12]    0.316804 dtype: float64 [2, 5]     0.263718 [5, 9]     0.389621 [9, 12]    0.346661 type: float64

I grouped ages into 3 groups: 3-5, 6-9 and 10-12 y.o. Among those whose parents got divorced: 

- for 3-5 y.o. the frequency was 1356 (31.86%)

- for 6-9 y.o. the frequency was 1551 (36.451%)

- for 10-12 y.o. the frequency was 1348 (31.68%)

Meaning that divorces in families with children in all three age groups are evenly distributed.

Among those whose parents died:

- for 3-5 y.o. the frequency was 620 (26.37%)

- for 6-9 y.o. the frequency was 916 (38.96%)

- for 10-12 y.o. the frequency was 815 (34.66%)

Meaning that  3-5 y.o. children whose parents died underrepresented  with about 10% difference in our sample.

[purepuremint]

Assignment: Running Your First Program

1. My program

import pandas as pd import numpy as nu

# bug fix for display formats to avoid run time errors - put after code for loading data above #pd.set_option('display.float_format', lambda x:'%f'%x)

data = pd.read_csv('nesarc_pds.csv', low_memory=False)

#upper-case all DataFrame column names  data.columns = map(str.upper, data.columns)

print (len(data)) #number of observations (rows) print (len(data.columns)) # number of variables (columns)

#setting variables you will be working with to numeric

data['S1Q2D'] = data['S1Q2D'].convert_objects(convert_numeric=True) data['S1Q2E'] = data['S1Q2E'].convert_objects(convert_numeric=True) data['S6Q1'] = data['S6Q1'].convert_objects(convert_numeric=True)

#counts and percentages (i.e. frequency distributions) for each variable print ('COUNTS FOR BIOLOGICAL OR ADOPTIVE PARENTS GET DIVORCED OR PERMANENTLY STOP LIVING TOGETHER') cdiv = data['S1Q2D'].value_counts(sort=False) print (cdiv)

print ('PERCENTAGES FOR BIOLOGICAL OR ADOPTIVE PARENTS GET DIVORCED OR PERMANENTLY STOP LIVING TOGETHER') pdiv = data['S1Q2D'].value_counts(sort=False, normalize=True) print (pdiv)

print ('COUNTS FOR HAD PANIC ATTACK') cpan = data['S6Q1'].value_counts(sort=False) print (cpan)

print ('PERCENTAGES FOR HAD PANIC ATTACK') ppan = data['S6Q1'].value_counts(sort=False, normalize=True) print (ppan)

#subset data to kids who had BIOLOGICAL OR ADOPTIVE PARENTS GOT DIVORCED OR PERMANENTLY STOP LIVING TOGETHER WHEN RESPONDENT WAS age 3 to 12 subdiv=data[(data['S1Q2E']>=3) & (data['S1Q2E']<=12) & (data['S1Q2D']==1)]

#make a copy of my new subsetted data subdiv2 = subdiv.copy()

#frequency distritions on new data frames print ('COUNTS FOR AGE 3 TO 12 y.o.') cage = subdiv2['S1Q2E'].value_counts(sort=False) print (cage)

print ('PERCENTAGES FOR AGE 3 TO 12 y.o.') page = subdiv2['S1Q2E'].value_counts(sort=False, normalize=True) print (page)

print ('COUNTS FOR DIVORCE') cdiv2 = subdiv2['S1Q2D'].value_counts(sort=False) print (cdiv2)

print ('PERCENTAGES FOR DIVORCE') pdiv2 = subdiv2['S1Q2D'].value_counts(sort=False, normalize=True) print (pdiv2)

2. The output that displays three of my variables as frequency tables + description 

COUNTS FOR BIOLOGICAL OR ADOPTIVE PARENTS GET DIVORCED OR PERMANENTLY STOP LIVING TOGETHER BEFORE RESPONDENT WAS 18 1     6914 2    30261 9       65 Name: S1Q2D, dtype: int64 PERCENTAGES FOR BIOLOGICAL OR ADOPTIVE PARENTS GET DIVORCED OR PERMANENTLY STOP LIVING TOGETHER 

1    0.160444 2    0.702225 9    0.001508 Name: S1Q2D, dtype: float64

Description: 

6914 or 0.160444 (16%) of the respondents whose caregivers got divorced before respondent was 18,

30261 or 0.702225 (70.2%) of the respondents had not and 65 or 0.001508 (0.15%) of the respondents did not answer.

COUNTS FOR HAD PANIC ATTACK 1     5278 2    36429 9     1386 Name: S6Q1, dtype: int64

PERCENTAGES FOR HAD PANIC ATTACK 1    0.122479 2    0.845358 9    0.032163 Name: S6Q1, dtype: float64

Description:

5278 or  0.122479 (12.2%) of the respondents had a panic attack or an anxiety episode at least ones in their life, 36429 or 0.845358 (84.5%) of the respondents had not, and 1386 or 0.032163 (0.32%) of the respondents did not answer.

COUNTS FOR AGE 3 TO 12 y.o.

3     420 4     374 5     562 6     406 7     388 8     422 9     335 10    478 11    332 12    538 Name: S1Q2E, dtype: int64

Description: Here is the amount of 3 to 12 years old respondents whose caregivers got divorced  (for each age group).

PERCENTAGES FOR AGE 3 TO 12 y.o. 3     0.098707 4     0.087897 5     0.132080 6     0.095417 7     0.091187 8     0.099177 9     0.078731 10    0.112338 11    0.078026 12    0.126439 Name: S1Q2E, dtype: float64

Description: Here is the distribution (percentages) of 3 to 12 years old respondents whose caregivers got divorced (for each age group).

COUNTS FOR DIVORCE 1    4255 Name: S1Q2D, dtype: int64

Description: Total amount of  3 to 12 years old respondents whose caregivers got divorced.

PERCENTAGES FOR DIVORCE 1    1 Name: S1Q2D, dtype: float64

Description: It shows that 100% of our sample are respondents whose caregivers got divorced.

[purepuremint]

Assignment: Getting Your Research Project Started

Data Management and Visualisation

After looking through the codebook for the NESARC study, I guess it would be interesting to look at the relationship between panic disorder/anxiety and childhood events and circumstances, such as lost of a parent or divorce.

Data set: NESARC

Literature review:

 The     study of  Takeshi Otowa and     colleagues (http://www.psy-journal.com/article/S0165-1781(14)00632-5/pdf) suggests that parental     separation was associated with a wide range of adult psychopathology,     whereas parental death was specifically associated with phobia and alcohol     dependence. Maternal and paternal separations were almost equally     associated with most forms of psychopathology. Structural equation     modeling suggested that parental loss accounted for about 10% of the     variance of adult psychopathology, of which parental separation had the     strongest impacts on risk for depression and drug abuse/dependence (11% of     the total variance). Their findings suggest that early parental     separation has stronger and wider effects on adult psychopathology than     parental death.

Kathryn H. Howell and     colleagues research (http://www.tandfonline.com/doi/full/10.1080/07481187.2014.951497) examined factors theorized to     contribute to adaptive functioning in 56 parentally bereaved children (age     7–13) who had lost their caregiver within the previous 6     months.  Linear mixed modeling revealed that children in the     adaptive functioning group had lower mean scores on avoidant coping and     higher mean scores on coping efficacy, religiosity, parental positive     reinforcement, and parental empathy. Findings suggest that adaptive     functioning following parental loss is related to both child-intrinsic     factors and child-extrinsic factors.

Mc Elroy S and Hevey D     study (http://www.ncbi.nlm.nih.gov/pubmed/24011494)  examined a relationship between adverse     early experiences, stressors, psychosocial resources and wellbeing. They     suggest that all early experiences, except physical abuse     and death of a parent in childhood, were significantly correlated     with increased number of, stressors and lower wellbeing scores. 

 The literature research helped me clarify my research question to formulate the hypothesis.

Hypothesis: There might be a grater correlation between panic disorder/anxiety and separation with a parent in early age due to divorce than the same disorder and separation in early age due to paren’s death.

Variables from the NESARC dataset that I will use: 

S1Q2D -  DID BIOLOGICAL OR ADOPTIVE PARENTS GET     DIVORCED OR PERMANENTLY STOP LIVING TOGETHER BEFORE RESPONDENT WAS 18

S1Q2E - AGE WHEN BIOLOGICAL/ADOPTIVE     PARENTS STOPPED LIVING TOGETHER

S6Q1  - HAD PANIC ATTACK, SUDDENLY FELT FRIGHTENED/OVERWHELMED/NERVOUS AS IF IN GREAT DANGER BUT WERE NOT

S1Q2K - DID BIOLOGICAL OR ADOPTIVE PARENT DIE BEFORE RESPONDENT WAS 18

S1Q2L - AGE AT DEATH OF BIOLOGICAL OR ADOPTIVE PARENT