Python Data Analysis for Healthcare - Note 2

Author

SEOYEON CHOI

Published

June 5, 2025

1

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
df_pharmacy_sales = pd.read_csv("../../../delete/Demo Pharmacy Sales Data.csv")
df_pharmacy_sales
Date Sold Med_name Med_class Quantity Sold Price
0 2021-05-07 Clotrimazole Topical Cream (2%) Antifungal 66 86.9
1 2021-08-09 Alprostadil Urethral Suppository (125 mcg) Prostaglandin E1 Analog 15 22.9
2 2021-06-15 Methyltestosterone Tablet (10 mg) Androgen Hormone 5 5.9
3 2021-02-19 Buspirone Tablet (5 mg) Anxiolytic 89 55.7
4 2022-09-24 Hydrocodone/Acetaminophen Tablet (5/325 mg) Opioid Analgesic/Analgesic Combination 79 0.7
... ... ... ... ... ...
999995 2020-11-29 Alprostadil Urethral Suppository (125 mcg) Prostaglandin E1 Analog 34 58.0
999996 2021-03-30 Fenoprofen Tablet (600 mg) Nonsteroidal Anti-Inflammatory Drug 12 98.3
999997 2020-04-17 Doxazosin Tablet (1 mg) Alpha-Blocker 83 10.3
999998 2021-12-08 Flumazenil Injection (0.1 mg/mL) Benzodiazepine Antagonist 1 23.9
999999 2022-08-16 Prednisolone Tablet (5 mg) Corticosteroid 44 54.5

1000000 rows × 5 columns

기본 정보 보기

df_pharmacy_sales.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1000000 entries, 0 to 999999
Data columns (total 5 columns):
 #   Column         Non-Null Count    Dtype  
---  ------         --------------    -----  
 0   Date Sold      1000000 non-null  object 
 1   Med_name       1000000 non-null  object 
 2   Med_class      1000000 non-null  object 
 3   Quantity Sold  1000000 non-null  int64  
 4   Price          1000000 non-null  float64
dtypes: float64(1), int64(1), object(3)
memory usage: 38.1+ MB

이 중 QUantity Sold랑 Price 만 숫자형인 것을 확인할 수 있음

summary_stats = df_pharmacy_sales[['Quantity Sold','Price']].describe()
summary_stats
Quantity Sold Price
count 1000000.000000 1000000.000000
mean 50.524566 50.024411
std 28.847235 28.872706
min 1.000000 0.100000
25% 26.000000 25.000000
50% 51.000000 50.000000
75% 75.000000 75.100000
max 100.000000 100.000000

summary 보면 위와 같음

df_pharmacy_sales['Quantity Sold'].plot.hist(bins=30,edgecolor='k')
plt.title("Histogram of Quantity Sold")
plt.show()

bin 을 어떻게 설정하느냐에 따라 약 판매량 데이터를 다르게 볼 수 있을듯

plt.figure(figsize=(12,4))
plt.subplot(1,2,1)
df_pharmacy_sales['Quantity Sold'].plot.hist(bins=20,edgecolor='k')
plt.title("Histogram of Quantity Sold")

plt.subplot(1,2,2)
df_pharmacy_sales['Price'].plot.hist(bins=20,edgecolor='k')
plt.title("Histogram of Price")

plt.show()

팔린 개수랑 가격이랑 어떻게 범위가 같지..

plt.figure(figsize=(12,4))
plt.subplot(1,2,1)
sns.histplot(df_pharmacy_sales['Quantity Sold'],bins=20,kde=True,color='blue')
plt.title("Histogram of Quantity Sold")

plt.subplot(1,2,2)
sns.histplot(df_pharmacy_sales['Price'],bins=20,kde=True,color='pink')
plt.title("Histogram of Price")

plt.show()

seaborn을 사용해서 데이터를 그려보면 위와 같은데

plot 그리는 방법은 다양하니 참고

2

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from wordcloud import WordCloud
df_pharmacy_sales = pd.read_csv("../../../delete/Demo Pharmacy Sales Data.csv")
df_pharmacy_sales.head()
Date Sold Med_name Med_class Quantity Sold Price
0 2021-05-07 Clotrimazole Topical Cream (2%) Antifungal 66 86.9
1 2021-08-09 Alprostadil Urethral Suppository (125 mcg) Prostaglandin E1 Analog 15 22.9
2 2021-06-15 Methyltestosterone Tablet (10 mg) Androgen Hormone 5 5.9
3 2021-02-19 Buspirone Tablet (5 mg) Anxiolytic 89 55.7
4 2022-09-24 Hydrocodone/Acetaminophen Tablet (5/325 mg) Opioid Analgesic/Analgesic Combination 79 0.7

단어 시각화를 위해 WordCloud를 쓸거임

df_pharmacy_sales['Med_name'].value_counts()
Tretinoin Topical Cream (0.025%)                     15027
Ketoconazole Topical Cream (2%)                      11008
Adapalene/Benzoyl Peroxide Topical Gel (0.1/2.5%)    10097
Triamcinolone Topical Ointment (0.1%)                10083
Clobetasol Topical Cream (0.05%)                      9984
                                                     ...  
Guanfacine Tablet (2 mg)                               939
Olopatadine Nasal Spray (665 mcg/spray)                938
Methylphenidate Tablet (10 mg)                         938
Lithium Carbonate Capsule (300 mg)                     920
Clindamycin Topical Lotion (1%)                        917
Name: Med_name, Length: 279, dtype: int64

Med_name의 summary는 위와 같음

len(df_pharmacy_sales.query('Med_name=="Guanfacine Tablet (2 mg)"'))
939
df_pharmacy_sales['Med_class'].value_counts()
 Nonsteroidal Anti-Inflammatory Drug                        82146
 Phosphodiesterase Type 5 Inhibitor                         56294
 Alpha-Blocker                                              52208
 Anticonvulsant                                             49766
 Beta-Blocker                                               40673
                                                            ...  
 Norepinephrine Reuptake Inhibitor                            995
 Melatonin Receptor Agonist                                   983
 Cardiac Glycoside                                            981
 Anticholinergic/Short-Acting Beta-2 Agonist Combination      977
 Mood Stabilizer                                              920
Name: Med_class, Length: 87, dtype: int64

Med_class의 summary는 위와 같음

top_number = 20
top_classes = df_pharmacy_sales['Med_class'].value_counts().nlargest(top_number).index
top_classes
Index([' Nonsteroidal Anti-Inflammatory Drug',
       ' Phosphodiesterase Type 5 Inhibitor', ' Alpha-Blocker',
       ' Anticonvulsant', ' Beta-Blocker', ' Antifungal', ' Corticosteroid',
       ' Benzodiazepine', ' Low-Potency Corticosteroid',
       ' High-Potency Corticosteroid', ' Opioid Analgesic', ' Antibiotic',
       ' Inhaled Corticosteroid', ' 5-Alpha Reductase Inhibitor', ' Analgesic',
       ' Prostaglandin E1 Analog', ' Retinoid', ' Sympathomimetic',
       ' Vasodilator', ' Alpha-2 Agonist'],
      dtype='object')
df_top_med_classes = df_pharmacy_sales[df_pharmacy_sales['Med_class'].isin(top_classes)]
df_top_med_classes
Date Sold Med_name Med_class Quantity Sold Price
0 2021-05-07 Clotrimazole Topical Cream (2%) Antifungal 66 86.9
1 2021-08-09 Alprostadil Urethral Suppository (125 mcg) Prostaglandin E1 Analog 15 22.9
6 2018-10-22 Norepinephrine Injection (2 mg/mL) Sympathomimetic 96 80.6
8 2022-05-19 Rofecoxib Tablet (25 mg) Nonsteroidal Anti-Inflammatory Drug 32 21.2
9 2019-07-21 Fluticasone Inhaler (50 mcg/actuation) Inhaled Corticosteroid 71 9.6
... ... ... ... ... ...
999992 2018-10-31 Carbamazepine Tablet (200 mg) Anticonvulsant 92 73.3
999995 2020-11-29 Alprostadil Urethral Suppository (125 mcg) Prostaglandin E1 Analog 34 58.0
999996 2021-03-30 Fenoprofen Tablet (600 mg) Nonsteroidal Anti-Inflammatory Drug 12 98.3
999997 2020-04-17 Doxazosin Tablet (1 mg) Alpha-Blocker 83 10.3
999999 2022-08-16 Prednisolone Tablet (5 mg) Corticosteroid 44 54.5

661970 rows × 5 columns

상위 20개만 살펴보고자 할때

df_top_med_classes = df_top_med_classes.groupby('Med_class')['Quantity Sold'].sum().reset_index()
df_top_med_classes
Med_class Quantity Sold
0 5-Alpha Reductase Inhibitor 1101549
1 Alpha-2 Agonist 802378
2 Alpha-Blocker 2625280
3 Analgesic 1061008
4 Antibiotic 1259910
5 Anticonvulsant 2514863
6 Antifungal 1819102
7 Benzodiazepine 1705063
8 Beta-Blocker 2062362
9 Corticosteroid 1776749
10 High-Potency Corticosteroid 1509454
11 Inhaled Corticosteroid 1112097
12 Low-Potency Corticosteroid 1519123
13 Nonsteroidal Anti-Inflammatory Drug 4153538
14 Opioid Analgesic 1514248
15 Phosphodiesterase Type 5 Inhibitor 2838613
16 Prostaglandin E1 Analog 1060112
17 Retinoid 1019449
18 Sympathomimetic 1005189
19 Vasodilator 973025 
type(df_top_med_classes.groupby('Med_class')['Quantity Sold'].sum())
pandas.core.series.Series

reset_index하기 전

type(df_top_med_classes)
pandas.core.frame.DataFrame

reset_index 후

df_top_med_classes = df_top_med_classes.sort_values(by='Quantity Sold')
df_top_med_classes
Med_class Quantity Sold
1 Alpha-2 Agonist 802378
19 Vasodilator 973025
18 Sympathomimetic 1005189
17 Retinoid 1019449
16 Prostaglandin E1 Analog 1060112
3 Analgesic 1061008
0 5-Alpha Reductase Inhibitor 1101549
11 Inhaled Corticosteroid 1112097
4 Antibiotic 1259910
10 High-Potency Corticosteroid 1509454
14 Opioid Analgesic 1514248
12 Low-Potency Corticosteroid 1519123
7 Benzodiazepine 1705063
9 Corticosteroid 1776749
6 Antifungal 1819102
8 Beta-Blocker 2062362
5 Anticonvulsant 2514863
2 Alpha-Blocker 2625280
15 Phosphodiesterase Type 5 Inhibitor 2838613
13 Nonsteroidal Anti-Inflammatory Drug 4153538

4

plt.barh(df_top_med_classes['Med_class'],df_top_med_classes['Quantity Sold'])
plt.title(f'Top {top_number} Medication Classes by Quantity Sold')
plt.xlabel("Quantity Sold")
plt.ylabel("Medication Class")
plt.show()

med_classes = df_pharmacy_sales['Med_class'].str.cat(sep='')
wordcloud = WordCloud(width=800,height=400).generate(med_classes)
plt.figure(figsize=(14,7))
plt.imshow(wordcloud,interpolation='bilinear')
plt.show()

med_names = df_pharmacy_sales['Med_name'].str.cat(sep='')
wordcloud = WordCloud(width=800,height=400).generate(med_names)
plt.figure(figsize=(10,5))
plt.imshow(wordcloud,interpolation='bilinear')
plt.show()

import pandas as pd
import matplotlib.pyplot as plt
import prophet
from prophet import Prophet
df_pharmacy_sales = pd.read_csv("../../../delete/Demo Pharmacy Sales Data.csv")
df_pharmacy_sales
Date Sold Med_name Med_class Quantity Sold Price
0 2021-05-07 Clotrimazole Topical Cream (2%) Antifungal 66 86.9
1 2021-08-09 Alprostadil Urethral Suppository (125 mcg) Prostaglandin E1 Analog 15 22.9
2 2021-06-15 Methyltestosterone Tablet (10 mg) Androgen Hormone 5 5.9
3 2021-02-19 Buspirone Tablet (5 mg) Anxiolytic 89 55.7
4 2022-09-24 Hydrocodone/Acetaminophen Tablet (5/325 mg) Opioid Analgesic/Analgesic Combination 79 0.7
... ... ... ... ... ...
999995 2020-11-29 Alprostadil Urethral Suppository (125 mcg) Prostaglandin E1 Analog 34 58.0
999996 2021-03-30 Fenoprofen Tablet (600 mg) Nonsteroidal Anti-Inflammatory Drug 12 98.3
999997 2020-04-17 Doxazosin Tablet (1 mg) Alpha-Blocker 83 10.3
999998 2021-12-08 Flumazenil Injection (0.1 mg/mL) Benzodiazepine Antagonist 1 23.9
999999 2022-08-16 Prednisolone Tablet (5 mg) Corticosteroid 44 54.5

1000000 rows × 5 columns

df_pharmacy_sales[df_pharmacy_sales['Med_class'].str.contains('Anxiolytic')][['Date Sold', 'Quantity Sold']]
Date Sold Quantity Sold
3 2021-02-19 89
122 2022-03-01 82
143 2020-08-11 17
327 2020-12-08 86
518 2019-03-31 70
... ... ...
999594 2019-08-11 42
999711 2020-05-16 71
999772 2019-09-02 63
999883 2019-09-09 38
999929 2023-03-25 4

10004 rows × 2 columns

Anxiolytic을 포함하는 문자열만 불러와서 Date Sole, Quantity Sold 변수만 보기

df_prophet = df_pharmacy_sales[df_pharmacy_sales['Med_class'].str.contains('Anxiolytic')][['Date Sold', 'Quantity Sold']]
df_prophet = df_prophet.reset_index()
df_prophet
index Date Sold Quantity Sold
0 3 2021-02-19 89
1 122 2022-03-01 82
2 143 2020-08-11 17
3 327 2020-12-08 86
4 518 2019-03-31 70
... ... ... ...
9999 999594 2019-08-11 42
10000 999711 2020-05-16 71
10001 999772 2019-09-02 63
10002 999883 2019-09-09 38
10003 999929 2023-03-25 4

10004 rows × 3 columns

df_prophet = df_prophet.rename(columns={'Date Sold': 'ds','Quantity Sold':'y'})
df_prophet
index ds y
0 3 2021-02-19 89
1 122 2022-03-01 82
2 143 2020-08-11 17
3 327 2020-12-08 86
4 518 2019-03-31 70
... ... ... ...
9999 999594 2019-08-11 42
10000 999711 2020-05-16 71
10001 999772 2019-09-02 63
10002 999883 2019-09-09 38
10003 999929 2023-03-25 4

10004 rows × 3 columns

type(df_prophet['ds'][0])
str
df_prophet['ds'] = pd.to_datetime(df_prophet['ds'])
type(df_prophet['ds'][0])
pandas._libs.tslibs.timestamps.Timestamp

문자형이었던 데이터 타입을 판다스 내부 타임스템프형으로 바꿔줌

model = Prophet()
model.fit(df_prophet)
21:18:56 - cmdstanpy - INFO - Chain [1] start processing
21:18:56 - cmdstanpy - INFO - Chain [1] done processing
<prophet.forecaster.Prophet at 0x7fcce5530fa0>
future = model.make_future_dataframe(periods=365)
forecast = model.predict(future)

365일 날짜를 생성해서 분석 수행하도록 periods를 365로 넣음

forecast.head()
ds trend yhat_lower yhat_upper trend_lower trend_upper additive_terms additive_terms_lower additive_terms_upper weekly weekly_lower weekly_upper yearly yearly_lower yearly_upper multiplicative_terms multiplicative_terms_lower multiplicative_terms_upper yhat
0 2018-10-20 50.088200 12.457751 86.071166 50.088200 50.088200 -0.101479 -0.101479 -0.101479 -0.559619 -0.559619 -0.559619 0.458139 0.458139 0.458139 0.0 0.0 0.0 49.986721
1 2018-10-21 50.089255 12.637593 84.826632 50.089255 50.089255 -1.108192 -1.108192 -1.108192 -1.314963 -1.314963 -1.314963 0.206771 0.206771 0.206771 0.0 0.0 0.0 48.981063
2 2018-10-22 50.090309 14.975067 86.555339 50.090309 50.090309 0.855300 0.855300 0.855300 0.904402 0.904402 0.904402 -0.049102 -0.049102 -0.049102 0.0 0.0 0.0 50.945609
3 2018-10-23 50.091364 11.951660 87.639434 50.091364 50.091364 0.347593 0.347593 0.347593 0.651931 0.651931 0.651931 -0.304338 -0.304338 -0.304338 0.0 0.0 0.0 50.438957
4 2018-10-24 50.092418 14.286608 83.483818 50.092418 50.092418 -0.474225 -0.474225 -0.474225 0.079648 0.079648 0.079648 -0.553873 -0.553873 -0.553873 0.0 0.0 0.0 49.618193 
type(forecast)
pandas.core.frame.DataFrame
fig = model.plot(forecast)
plt.show()

fig = model.plot_components(forecast)
plt.show()

  1. trend로 장기적인 경향을 시각화

  2. Weekly Seasonality (주간 계절성) 요일별 주기적인 패턴 확인

  3. Yearly Seasonality (연간 계절성) 월 또는 계절 단위의 반복되는 패턴을 시각화