-
Notifications
You must be signed in to change notification settings - Fork 0
/
📈Data_Viewer.py
216 lines (191 loc) · 8.94 KB
/
📈Data_Viewer.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
#zebrafish monitor data viewer
import streamlit as st
import pandas as pd
import numpy as np
from datetime import timedelta
import plotly.express as px
from plotly.subplots import make_subplots
import plotly.graph_objects as go
from statsmodels.tsa.seasonal import seasonal_decompose
from statsmodels.tsa.stattools import adfuller
from statsmodels.graphics.tsaplots import plot_acf, plot_pacf
from statsmodels.tsa.statespace.sarimax import SARIMAX
from statsmodels.tools.eval_measures import rmse
import datetime
st.set_page_config(page_title="Zebrafish Activity Data Viewer", layout="wide")
st.write("""<style>
div.block-container{padding-top:0rem;}
div[data-testid="stMetricValue"] > div {font-size: 0.6rem;}
</style>""", unsafe_allow_html = True)
st.title("Zebrafish Activity Data Viewer")
st.info("""~ Welcome to the demonstration of time series data analysis for Zebrafish Activity Monitor System.
Please use the left side bar to select data and parameters for analysis.""", icon="🐟")
#display some info at the bottom of side bar
def site_info():
st.markdown("---")
github = "[](https://github.com/alexkychen/zebrafish-monitor)"
st.markdown(github)
st.markdown("Developer: Alex Chen ([Contact me](mailto:alexkychen@gmail.com))")
st.markdown("Copyright © 2023")
@st.cache
def read_data():
df = pd.read_csv("data/data8.csv", index_col=0)
df.index = pd.to_datetime(df.index)
#calculate total count for 4 sensors
df["Total"] = df.iloc[:].sum(axis=1).astype(int)
first_date = df.index.min()#.strftime("%Y-%m-%d")
last_date = df.index.max()#strftime("%Y-%m-%d")
return df, first_date, last_date
#create sidebar
def sidebar_param(options, first_date, last_date):
"""
options: df's column names
first_date: first datetime index from df
last_date: last datetime index from df
"""
with st.sidebar:
choices = st.multiselect("Select sensor to display", options, default="Total")
#select date range
start_date = st.date_input("Select start date", value=first_date, min_value=first_date, max_value=last_date)
end_date = st.date_input("Select end date", value=first_date+timedelta(days=10), min_value=first_date, max_value=last_date)
#select resample frequency (mean count for every X minutes)
freq = st.number_input("Resample frequency (minute) for mean count", min_value=1, max_value=1440, value=15)
st.markdown("Show feeding time on main plot")
vlines = [False, False, False]
vlines[0] = st.checkbox("9 am")
vlines[1] = st.checkbox("12 pm")
vlines[2] = st.checkbox("16 pm")
site_info()
return choices, start_date, end_date, freq, vlines
#subset data
def subset_data(df, start_date, end_date, freq):
#convert datetime to string
start = start_date.strftime("%Y-%m-%d")
end = end_date.strftime("%Y-%m-%d")
#subset data
df = df[(df.index >= start) & (df.index <= end)]
#resample for mean count
period = str(freq)+"T"
df = df.resample(period).mean()
df.interpolate(method="linear", inplace=True)
return df, start, end, period
#create main plot
def main_plot(df, freq, choices, vlines, start, end):
#create Y label text
if freq == 1:
ylab = "Count per minute"
else:
ylab = "Mean count of every " + str(freq)+ " minutes"
#create plotly plot object
fig = px.line(df, y=choices, labels={"value":ylab, "datetime":"Date/time","variable":"Sensor"})
#Add vertical line to indicate feeding time
if vlines[0]:
for i in pd.date_range(start, end):
TL = i.strftime("%Y-%m-%d") + " 09:00"
fig.add_vline(x=TL, line_color="gray")
if vlines[1]:
for i in pd.date_range(start, end):
TL = i.strftime("%Y-%m-%d") + " 12:00"
fig.add_vline(x=TL, line_color="gray")
if vlines[2]:
for i in pd.date_range(start, end):
TL = i.strftime("%Y-%m-%d") + " 16:00"
fig.add_vline(x=TL, line_color="gray")
st.plotly_chart(fig, use_container_width=True)
#ETS decomposition
def ets_decomposition(df, choices, freq):
co1, co2 = st.columns((1,4))
co1.subheader("ETS decomposition")
selected_data = co1.selectbox("Select sensor data", choices)
selected_model = co1.selectbox("Select a method", ["additive", "multiplicative"])
calculated_period = int(1440 / freq)
co1.markdown("Period: **"+str(calculated_period)+"**")
co1.caption("Period = 1440 / resample freq (min)")
#co1.metric(label="Period",value=calculated_period, help="Period = 1440 / freq (min)")
try:
results_add = seasonal_decompose(df[selected_data], model=selected_model, period=calculated_period)
#create ETS decomposition plot
fig = make_subplots(rows=4, cols=1)
fig.add_trace(go.Scatter(x=results_add.observed.index, y=results_add.observed), row=1, col=1)
fig.add_trace(go.Scatter(x=results_add.trend.index, y=results_add.trend), row=2, col=1)
fig.add_trace(go.Scatter(x=results_add.seasonal.index, y=results_add.seasonal), row=3, col=1)
fig.add_trace(go.Scatter(x=results_add.resid.index, y=results_add.resid, mode='markers'), row=4, col=1)
fig.add_hline(y=0, row=4, col=1)
fig.update_yaxes(title_text="Observed", row=1, col=1)
fig.update_yaxes(title_text="Trend", row=2, col=1)
fig.update_yaxes(title_text="Seasonal", row=3, col=1)
fig.update_yaxes(title_text="Residual", row=4, col=1)
fig.update_layout(height=600, showlegend=False, margin=dict(t=35))
co2.plotly_chart(fig, use_container_width=True)
#st.plotly_chart(results_add.plot())
except:
co2.error("Error: Can't generate plots. Please select data or adjust parameters (e.g., Resample frequency).")
return selected_data
#ACF and PACF
def acf_pacf(df, selected_data):
co1, co2, co3 = st.columns((1,2,2))
co1.subheader("ACF and PACF")
selected_lags = co1.number_input("Enter max. lags to plot", min_value=10, max_value=250, value=100)
co1.caption("Y-axis: Correlation coefficient")
co1.caption("X-axis: Lag number")
try:
co2.pyplot(plot_acf(df[selected_data], lags=selected_lags))
except:
co2.error("Error: Can't generate ACF plot. Please re-select data or adjust parameters")
try:
co3.pyplot(plot_pacf(df[selected_data], lags=selected_lags, method="ywm"))
except:
co3.error("Error: Can't generate PACF plot. Please re-select data or adjust parameters")
#perform ADF stationarity test
def adf_test(df, selected_data):
st.subheader("Augmented Dicky-Fuller Test")
#st.markdown("Null hypothesis: time series has a unit root.")
try:
adftest = adfuller(df[selected_data], autolag="AIC")
except:
st.error("Error: Unable to perform ADF test. Please re-select data or adjust parameters.")
st.stop()
teststats, pvalue, lags, nobs = format(adftest[0], ".5g"), format(adftest[1], ".4g"), adftest[2], adftest[3]
crit1, crit5, crit10 = adftest[4]["1%"], adftest[4]["5%"], adftest[4]["10%"]
co1, co2, co3 = st.columns((1,1,2))
co1.markdown("""
- Test statistic: **""" + str(teststats) + """**
- *p*-value: **""" + str(pvalue) + """**
- Number of lags used: **""" + str(lags) + """**
- Number of observations: **""" + str(nobs) + """**
""")
co2.markdown("""
- Critical value (1%): **""" + str(format(crit1, ".5g"))+ """**
- Critical value (5%): **""" + str(format(crit5, ".5g"))+ """**
- Critical value (10%): **""" + str(format(crit10, ".5g"))+ """**
""")
#make conclusions
co3.markdown("**Conclusion of data stationarity:**")
if float(pvalue) > 0.05:
co3.markdown("""*p*-value is greater than 0.05.
We can't reject null hypothesis, and time series is considered as non-stationary.
Applying differencing to time series is required for ARIMA based models.""")
else:
co3.markdown("""*p*-value is less than 0.05.
We can reject null hypothesis, and time series is considered as stationary.
Applying differencing to time series is not required for ARIMA based models.""")
#Predicted result comparison among models
def compare_result(filepath):
dfr = pd.read_csv(filepath, index_col=0)
model_options = dfr.columns[1:]
model_select = st.multiselect("Select models", model_options)
if __name__ == "__main__":
#read data
df, first_date, last_date = read_data()
#create sidebar and get some params
choices, start_date, end_date, freq, vlines = sidebar_param(df.columns, first_date, last_date)
#subset data
df, start, end, period = subset_data(df, start_date, end_date, freq)
#create main plot
main_plot(df, freq, choices, vlines, start, end)
#ETS decomposition
selected_data = ets_decomposition(df, choices, freq)
#ACF and PACF
acf_pacf(df, selected_data)
#ADF test
adf_test(df, selected_data)