Convertir columna de string a datetime
df['Date'] = pd.to_datetime(df['Date'], format='%Y-%m-%d')
Agrupar datos de por hora y hacer un boxplot de cada grupo
durations['Time'] = pd.to_datetime(durations.Time)
filtered = durations[(durations['Time'] > datetime.datetime(2021,4,19,16)) \
& (durations['Time'] < datetime.datetime(2021,4,21,16))]
groups = filtered.groupby(pd.Grouper(key="Time", freq="H"))
hours = pd.concat([pd.DataFrame(x[1].values) for x in groups], axis=1)
hours = pd.DataFrame(hours)
hours.columns = range(len(hours.columns))
hours.plot.box(figsize=(16, 10))