spc_plotly is a Python helper library for creating XmR Charts according to the theories of Statistical Process Control using the Plotly library.
XmR Charts allow the viewer to quickly identify signals in a data set and ignore routine variation.
pip install spc-plotlyfrom spc_plotly import xmr
import pandas as pd
counts = [
2478, 2350, 2485, 2296, 2359, 2567, 3089, 2668, 1788, 2854,
2365, 1883, 1959, 1927, 2640, 2626, 2144, 2409, 4412, 3287,
3049, 3364, 3078, 2972, 3415, 2753, 3102, 2191, 3693, 4385,
4699, 3031, 2659, 3885, 2628, 2621, 3071, 2002
]
periods = [
'2021-01', '2021-02', '2021-03', '2021-04', '2021-05', '2021-06',
'2021-07', '2021-08', '2021-09', '2021-10', '2021-11', '2021-12',
'2022-01', '2022-02', '2022-03', '2022-04', '2022-05', '2022-06',
'2022-07', '2022-08', '2022-09', '2022-10', '2022-11', '2022-12',
'2023-01', '2023-02', '2023-03', '2023-04', '2023-05', '2023-06',
'2023-07', '2023-08', '2023-09', '2023-10', '2023-11', '2023-12',
'2024-01', '2024-02'
]
data = pd.DataFrame({
"Period": periods,
"Count": counts
})
xmr_chart = xmr.XmR(
data=data,
x_ser_name="Period",
y_ser_name="Count",
x_cutoff="2023-06",
date_part_resolution="month", # This should match your data
custom_date_part="",
xmr_function="mean"
)
xmr_chart.mR_limit_values
# {
# 'mR_xmr_func': 571.8918918918919,
# 'mR_upper_limit': 1868.9427027027025,
# 'xmr_func': 'mean'
# }
xmr_chart.npl_limit_values
# {
# 'y_xmr_func': 2820.6315789473683,
# 'npl_upper_limit': 4341.864011379801,
# 'npl_lower_limit': 1299.3991465149359,
# 'xmr_func': 'mean'
# }
xmr_chart.signals
# {'anomalies': [(datetime.datetime(2022, 7, 1, 0, 0), 4412, 'High'),
# (datetime.datetime(2023, 6, 1, 0, 0), 4385, 'High'),
# (datetime.datetime(2023, 7, 1, 0, 0), 4699, 'High')],
# 'long_runs': [[('2021-11', 2365, 'Low'),
# ('2021-12', 1883, 'Low'),
# ('2022-01', 1959, 'Low'),
# ('2022-02', 1927, 'Low'),
# ('2022-03', 2640, 'Low'),
# ('2022-04', 2626, 'Low'),
# ('2022-05', 2144, 'Low'),
# ('2022-06', 2409, 'Low')]],
# 'short_runs': [[('2023-04', 2191, 'High'),
# ('2023-05', 3693, 'High'),
# ('2023-06', 4385, 'High'),
# ('2023-07', 4699, 'High'),
# ('2023-08', 3031, 'High')],
# [('2021-11', 2365, 'Low'),
# ('2021-12', 1883, 'Low'),
# ('2022-01', 1959, 'Low'),
# ('2022-02', 1927, 'Low'),
# ('2022-03', 2640, 'Low')]]}
xmr_chart.xmr_chart
For reference, please read Making Sense of Data by Donald Wheeler and Twenty Things You Need To Know
Some data naturally increases over time, e.g., price of some goods. When modeling this data, you should use sloped limit lines to identify signals.
xmr_chart = xmr.XmR(
data=data,
x_ser_name="Period",
y_ser_name="Count",
x_cutoff="2023-06",
xmr_function="mean"
sloped=True
)
If your data contains extreme outliers, you can update the xmr_function parameter to "median"
xmr_chart = xmr.XmR(
data=data,
x_ser_name="Period",
y_ser_name="Count",
x_cutoff="2023-06",
xmr_function="median"
sloped=False
)If you want to calculate the limits from a subset of your data, use the x_cutoff parameter.
xmr_chart = xmr.XmR(
data=data,
x_ser_name="Period",
y_ser_name="Count",
x_cutoff="2023-06",
xmr_function="median"
sloped=False
)This will include all data up to, and including, "2023-06". If no value is passed, the limits will be calculated using all data. Currently, this does not support an x_start, so I would recommend filtering the passed-in dataframe to start at the desired period.
Plotly, Pandas, and Numpy
