https://www.analyzemydata.org/archive daily 0.75 2023-05-16 https://www.analyzemydata.org/archive/2018/4/24/recurrent-neural-networks monthly 0.5 2018-04-26 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1524608428846-S6BQ6170UA5X9FK87DUV/fractal-cauliflower.jpg Archive - Recurrent Neural Networks https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1524609212892-2Q77E1OHNK6RVI7UU7GB/Slide02.jpg Archive - Recurrent Neural Networks Figure 1. Diagram of a two-layer deep neural network https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1524609604636-CCMSNASSJQSH559DFZQV/Slide03.jpg Archive - Recurrent Neural Networks Figure 2. Fitting a neural network. Adapted from https://www.embedded-vision.com/platinum-members/cadence/embedded-vision-training/documents/pages/neuralnetworksimagerecognition https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1524618554118-ZOII3GKB9O7O45VRQN97/Slide04.jpg Archive - Recurrent Neural Networks Figure 3. Analogy of deep learning model to visual recognition of a car. From https://www.slideshare.net/grigorysapunov/deep-learning-and-the-state-of-ai-2016. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1524619354174-DSSAFPZ9VL7N6DNHIYQQ/Slide06.jpg Archive - Recurrent Neural Networks Figure 4. Top: A single recurrent neuron, with input (x) and output (y); with unfolding over time. Bottom: Various configurations of recurrent neural networks, depending on the task. From the excellent blog: http://karpathy.github.io/2015/05/21/rnn-effectiveness/ https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1524619783183-YTEXCYTDT238TDYSSFCT/Slide07.jpg Archive - Recurrent Neural Networks Figure 5. A long short-term memory (LSTM) neuron. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1524745924883-M9OU7XTE8UTMUBL5MPWY/Fitbit_RNN%2C+summ_Page_4.jpg Archive - Recurrent Neural Networks Figure 6. Time series for each parameter from Fitbit data https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1524746383797-NG783MBXMNRBVHG3568X/Fitbit_RNN%2C+summ_Page_7.jpg Archive - Recurrent Neural Networks Figure 7. Training error over training epoch. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1524746462335-VM8X55CNJXDWQ1VN2BU4/Fitbit_RNN%2C+summ_Page_8.jpg Archive - Recurrent Neural Networks Figure 8. Actual (blue) and predicted (orange) sleep predicted by the model. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1524746892815-ATI6Y578V5VXWM28ZFVU/Fitbit_RNN_Multiple_lags_Page_7.jpg Archive - Recurrent Neural Networks Figure 9. Fitting a model with 7 days of lagged values. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1524746968086-BNA3N9PFFNPZIHPH7IWA/Fitbit_RNN_Multiple_lags_Page_8.jpg Archive - Recurrent Neural Networks Figure 10. Predicted minutes of sleep from model with 7 days of lagged data https://www.analyzemydata.org/archive/2017/11/2/machine-learning-teaser monthly 0.5 2017-11-16 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1510871232913-LVZOSVX53JS4GWKNIHD4/HK-Tank_HK-Aerial.jpg Archive - Machine Learning Teaser https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1510875624995-PUSBJ7LQTT2ECZ5ZBF89/gears-cogs-machine-machinery-159298.jpg Archive - Machine Learning Teaser https://www.analyzemydata.org/archive/2017/10/6/states-and-hmms monthly 0.5 2018-04-09 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1507752082533-YS70QYUIH91001M831H0/grasshopper-viridissima-green-corn-leaf-65642.jpg Archive - States and HMMs https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1507753102076-8YMWXT4OTYURC48DDYM8/HMM_number_of_awakenings_example_localdecode.jpeg Archive - States and HMMs Figure 1. Three-state HMM applied to number of awakenings from my Fitbit monitor. See prior post for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1507310853331-F99QMET3V6Z9RDSGGL3W/Steps_summGraph.jpeg Archive - States and HMMs https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1507310861273-8C0Z94P9Q33J3A5KM8TP/_histogram_Steps.jpeg Archive - States and HMMs https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1507576603600-QQCM5O54JWP3JN4IWVQE/HMM_steps2.jpeg Archive - States and HMMs Figure 3. 2-State HMM for Daily Steps. See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1507575297926-JI2H16WKTP2VP8OPGW6F/HMM_steps3.jpeg Archive - States and HMMs Figure 4. 3-state HMM for Daily Steps. See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1507664421313-H1LDTBARHR709H8TOR52/TSPlot_sleepGoal.jpeg Archive - States and HMMs Figure 5. Binary time series for sleep goal (0, 1). See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1507730141290-Q2W67IFJDM3C73GIRVYD/HMM_sleepGoal.jpeg Archive - States and HMMs Figure 6. HMM decoded plot for sleep goal. See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1507750987905-F3U1QABFPNJBXB64DHZW/TSPlot_exercise.jpeg Archive - States and HMMs https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1507751002629-4QE4Z8H2AE7PQSS8OZWM/Hist_exercise.jpeg Archive - States and HMMs https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1507750884996-PDG6CWPOZPKU5O9OTNMS/ZeroInfl_exercise_5state.jpeg Archive - States and HMMs Figure 8. Zero-inflated 5-state Poisson HMM, decoded with Viterbi algorithm. See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1507753739603-IC4G6FW19N0U3FY0TFVI/switzerland-zermatt-mountains-snow.jpg Archive - States and HMMs https://www.analyzemydata.org/archive/2017/9/13/hidden-markov-models monthly 0.5 2018-04-09 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1505394707557-5SEM3BINXMKNBT2RTB61/pexels-photo-416160.jpg Archive - Hidden Markov Models https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1505310871689-8EDOINVE2BMAZRXKQ47H/9781584885733.jpg Archive - Hidden Markov Models https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1505329269573-ZOXQTLDLFVGSPMPQ1ASG/540px-Gaussian-mixture-example.svg.png Archive - Hidden Markov Models Figure 1. Mixture distribution of three normal distributions with means of 5, 10, and 15. See text for details. From Wikipedia.org. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1505334250530-4UWWG820QOB6FCQL9QL4/Slide1.png Archive - Hidden Markov Models Figure 2. Hidden Markov Model with observations (X) generated from distributions (C), with transition probability matrix (Γ). See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1505391191195-MCLVRBQM67CSYDV3S75Y/HMM_number_of_awakenings_example_tsplot.jpeg Archive - Hidden Markov Models https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1505391200529-I4X5M9BYZD4J40OPHHJI/HMM_number_of_awakenings_example_histogram.jpeg Archive - Hidden Markov Models https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1505393163032-8ZS43YL3H6ZAGJ5H88OB/HMM_number_of_awakenings_example_localdecode.jpeg Archive - Hidden Markov Models Figure 4. Local decoding of three-state HMM. See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1505394015364-7ZLEKDZIE12PYSMIQSK4/HMM_number_of_awakenings_example_onestepforecast.jpeg Archive - Hidden Markov Models https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1505394032500-LD2OI40X3W11L1A3XHGW/HMM_number_of_awakenings_example_marginalvs50dayforecast.jpeg Archive - Hidden Markov Models https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1505394886604-35D2LSTXVRHCPBZB8HQ9/pexels-photo-57627.jpg Archive - Hidden Markov Models https://www.analyzemydata.org/archive/2017/8/23/seasonal-adjustment-part-ii monthly 0.5 2018-04-09 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503501638262-AHCUIKB108L4887HAVBZ/pexels-photo-262113.jpg Archive - Seasonal adjustment, Part II https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503938114562-Q4GDO8OKNTYTCCWM869E/acf_steps.jpeg Archive - Seasonal adjustment, Part II Figure 1. ACF (top) and PACF (bottom) plots of Fitbit step data for 100 lags. Peaks are seen in both plots are roughly every 7 to 14 days. See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503955991424-L5EM966GOEXWVAGSMJVI/image-asset.jpeg Archive - Seasonal adjustment, Part II Figure 2. Forecast plot from ARIMA(0,0,0) for steps for the next 30 days, with a fan chart confidence interval plot out to maximum 99%. See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503956837258-8D1MWR94T15ZWQQ8VSZL/model2Seasonal_steps.jpeg Archive - Seasonal adjustment, Part II Figure 3. The best seasonal ARIMA model with 30 day forecast based on seasonal loop. See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503957870171-6OA2TQ5I7ESMH8BYWRY0/periodogram5_steps.jpeg Archive - Seasonal adjustment, Part II https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503957895904-PNUQ3ZKO21YI8S71J30K/seasonalARIMA_steps.jpeg Archive - Seasonal adjustment, Part II https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503959716492-77NN8HQASL917MCGHU9Q/linearModel_steps.jpeg Archive - Seasonal adjustment, Part II https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503959733234-ZG7THXDCP7FC6ZZ89TB9/acfModel_steps.jpeg Archive - Seasonal adjustment, Part II https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503963497070-M6Q4AF24YCJIRLFR4VMO/linearModelLagACF_steps.jpeg Archive - Seasonal adjustment, Part II https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503963511989-CPIFS6KFFTEZPA24H33D/linearBestModel_steps.jpeg Archive - Seasonal adjustment, Part II https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503964047750-G717E85EZRBYG39H3FFU/road-asphalt-space-sky-56832.jpg Archive - Seasonal adjustment, Part II https://www.analyzemydata.org/archive/2017/7/26/seasonal-adjustment-part-1 monthly 0.5 2018-04-09 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503359794181-NXJ7L6VK39TZXGRI6CNE/pexels-photo-230629.jpg Archive - Seasonal adjustment, Part 1 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503333158790-VH014Z3RW8N5KJO669ZM/Cosine_curve_steps.jpeg Archive - Seasonal adjustment, Part 1 Figure 1. Linear regression fitted line (red) and daily steps from Fitbit. See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503333107081-H75KQDL7YV0TKCQJ0ZEL/Cosine_curves_2periods_steps.jpeg Archive - Seasonal adjustment, Part 1 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503349379825-HU3F8E66362LTOZWTYS9/Periodogram_steps.jpeg Archive - Seasonal adjustment, Part 1 Figure 3. Periodogram of steps created using fast Fourier transform. No kernel filters or tapering applied. See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503352479214-2JVQCEWKJ8IATD2H6IR2/Periodogram_smooth_steps.jpeg Archive - Seasonal adjustment, Part 1 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503352491101-915I6XRYA0BBFMAS7WVT/Periodogram_smooth2_steps.jpeg Archive - Seasonal adjustment, Part 1 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503352501380-TP9K0BN34891BV0L227M/Periodogram_smooth3_steps.jpeg Archive - Seasonal adjustment, Part 1 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503353477525-VSVUDBJEQ8WH0ZZP3IHM/Cosine_curves_3periods_steps.jpeg Archive - Seasonal adjustment, Part 1 Figure 5. Steps with additional periods included as determined by spectral analysis. See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503355092643-A9ZHYQCDAB4CV8XNYX2F/Periodogram_CI_steps.jpeg Archive - Seasonal adjustment, Part 1 Figure 6. Confidence intervals around the top two peaks obtained using convoluted Daniell kernel (m = (7,7)). Y-axis is log-scale. Lower peak is at frequency = 0.171 (red), and confidence interval is dashed red lines. Upper peak is at frequency = 0.435 (blue), and confidence intervals are in dashed blue. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1503362208174-DA031Y25R7OOPSGYCGQ9/maroonbells2.jpg Archive - Seasonal adjustment, Part 1 https://www.analyzemydata.org/archive/2017/6/14/seasonal-patterns-of-continuous-variable monthly 0.5 2018-04-09 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1498591626117-AUQPYFHQSKTW8HMPGG2H/image-asset.jpeg Archive - Seasonal patterns of continuous variable https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1498507558464-3PYVO3E2ZZEB8IEXOJLD/TS_example.jpg Archive - Seasonal patterns of continuous variable Figure 1. Hypothetical outcome with a cyclical pattern over time. Outlay shows the mistake of sampling without accounting for the seasonal pattern and incorrectly inferring a positive effect of an intervention. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1498508970191-SNY740FY413W0CR9GX9V/image-asset.jpeg Archive - Seasonal patterns of continuous variable Figure 2. Hypothetical outcomes measured over time. Both outcomes (blue and red lines) appear to be highly correlated due to cyclical nature of each. However, after 'adjusting' for the cyclical nature (inset), we see that there is no clear correlation between the two. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1498587943587-WA2RB13NS5JLWAT2XAGO/freqencyPlotSteps.jpeg Archive - Seasonal patterns of continuous variable - Periodogram Steps https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1498587983765-BAWL575BM3F9SYKPXABV/freqencyPlot.jpeg Archive - Seasonal patterns of continuous variable - Periodogram Sleep https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1498586884442-IMKHKH1ARG828TRZ9BZZ/dailySteps.jpg Archive - Seasonal patterns of continuous variable https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1498586897433-DXNY61S0C7VIYU56ALJ0/dailySleep.jpg Archive - Seasonal patterns of continuous variable https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1498586931355-18C5XBE6S0TQ0B2WNHRB/monthlySteps.jpg Archive - Seasonal patterns of continuous variable https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1498586942133-PJZ56M0BAGLH7ERHT120/biweeklySleep.jpg Archive - Seasonal patterns of continuous variable https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1498591579991-YD63TT71JO9O2EDDTBKX/image-asset.jpeg Archive - Seasonal patterns of continuous variable https://www.analyzemydata.org/archive/2017/5/5/correlation-between-two-continuous-measures monthly 0.5 2018-04-09 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1494018243884-6T21VZ53HUX8ROSD5OKS/image-asset.jpeg Archive - Correlation between two continuous measures https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1494003799111-0GV60YD3BQC2UK6TEAYQ/ThreeTSPlot.jpeg Archive - Correlation between two continuous measures Time series for three continuous variables, collected by my Fitbit device. Sleep (coded as 'Minutes.Asleep'), Steps, and Very Active (coded as 'Minutes.Very.Active'). The data has been filtered for days when less than 70% of a day was recorded. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1494007374651-7OVR11DPVYD1I0FL3BMI/ccfPlot.jpeg Archive - Correlation between two continuous measures Cross-correlation functions between activity and sleep. The lag at the bottom refers to the first variable listed--for example, in the first graph, the lag refers to the number of steps taken before or after the minutes of sleep was measured. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1494008979694-Q1SQ5B2PY3C7CENSUG8X/image-asset.jpeg Archive - Correlation between two continuous measures The lagged plot for Sleep based on lagged amounts of Steps. Created using the lag2.plot function of the astsa package in R. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1494011899638-MRG5HWP098LKAJ5M94D3/image-asset.jpeg Archive - Correlation between two continuous measures Predicted Minutes of Sleep after a given number of steps taken the day before based on our regression model. Dashed lines are 95% confidence intervals. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1494015616279-BK1FQ64NDRFW3RMRA589/image-asset.jpeg Archive - Correlation between two continuous measures https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1494016334229-FZ2LOGPHE8EFDCN0VTXG/image-asset.jpeg Archive - Correlation between two continuous measures Based on auto.arima, which selected a seasonal model with AR1 and SAR1 components around a period of 7. The astute person might note that the U-shaped quadratic relationship was not captured here, which is primarily due to a limit in the auto.arima function, which only allows a single xreg variable, rather than both the linear and quadratic predictors. https://www.analyzemydata.org/archive/2017/4/24/intervention-on-continuous-variable monthly 0.5 2018-04-09 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1493063473190-TYUX06UOJ2SEQ3Z5XSOI/Steps_summgraph.jpeg Archive - Intervention on continuous variable Example of daily steps obtained from a Fitbit device over a period of 5 months. Shown are the actual number of steps per day, the linear trend, the mean steps, and the lowess smooth (local trend). Code for this graph is available on Github. https://www.analyzemydata.org/archive/category/Intervention+Analyses monthly 0.5 https://www.analyzemydata.org/impressions daily 0.75 2018-10-23 https://www.analyzemydata.org/impressions/2018/10/23/individualized-risk-stratification-of-sudden-cardiac-death-a-new-take-on-components-of-the-decision-making-process monthly 0.5 2018-10-23 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1540302288457-XNAWDQMWBLNU6WSX52I4/cardiac-defibrillator-274986.jpg Impressions - Individualized Risk Stratification of Sudden Cardiac Death: A New Take on Components of the Decision-Making Process https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1540302491988-JXWMPID9WF26DFH0R71V/Individualized_Risk_strat%2C+Figure_1.jpg Impressions - Individualized Risk Stratification of Sudden Cardiac Death: A New Take on Components of the Decision-Making Process Figure 1. Overlapping components of decisions about sudden cardiac death and defibrillators. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1540309422321-R375KHZVX0HKRDSEUF00/Slide1.jpg Impressions - Individualized Risk Stratification of Sudden Cardiac Death: A New Take on Components of the Decision-Making Process https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1540308713111-9W27VQYEO65VHMGB16C6/Slide1.jpg Impressions - Individualized Risk Stratification of Sudden Cardiac Death: A New Take on Components of the Decision-Making Process https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1540308940875-CHA3YDWYSJPP99WBON5C/Individualized_Risk_strat%2C+Figure_2.jpg Impressions - Individualized Risk Stratification of Sudden Cardiac Death: A New Take on Components of the Decision-Making Process Figure 2. Perceived vs. actual gains and losses under Prospect theory. The ordinate is the ‘actual’ gain or loss as described in Scenario 1 (Box 2), and the abscissa is the perceived gain or loss (utility). Under Prospect theory, individuals tend to overweight perceived losses to gains. See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1540309202122-5ID6WEJWJR0RYP6KA9D0/Table_1.png Impressions - Individualized Risk Stratification of Sudden Cardiac Death: A New Take on Components of the Decision-Making Process https://www.analyzemydata.org/impressions/2018/7/15/5-feedback monthly 0.5 2018-08-08 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1533675249862-1BFSAJX4Y62L2PO5NYBK/feedback-2.jpg Impressions - 5. Feedback https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1533675545037-FQSCVKTNK6AIDTSA7Y6P/14946646966_51510ca3f0_b.jpg Impressions - 5. Feedback Figure 1. Local and Global optimum. Oftentimes we may be at a local optimum, in which case we actually need to get less optimal in order to reach the global optimum. See text for details. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1533677241938-KAP9R99VM4S4G5ARBWFR/cute.jpg Impressions - 5. Feedback https://www.analyzemydata.org/impressions/2018/7/15/4-analytics monthly 0.5 2018-08-08 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1531954770324-S52374EH66H2M2XK6L3G/Slide1.jpg Impressions - 4. Analytics Figure 1. Data Model for an individualized analysis approach. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1532366039407-Y3FTXDU9M3FZDFYFIXCZ/Slide1.jpg Impressions - 4. Analytics Figure 2. Example of Bayesian analysis applied to coin flip. In both examples, the prior probability distribution (top) favors a biased coin that winds up head 75% of the time (mode = 0.75). Initially (middle, left), after only 5 flips (2 heads), the posterior distribution (bottom) remains with a mode of 0.69 since it heavily favors the prior. However, after 500 flips (200 heads, middle, right), the posterior distribution favors tails with a mode of 0.415 for heads. https://www.analyzemydata.org/impressions/2018/7/15/3-clinical-application monthly 0.5 2018-08-08 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1533674520347-L6DCR4H9HBJ4WRK9U64A/doctors-stethoscope-500x500.jpg Impressions - 3. Clinical Application https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1531957040387-OJJEVMKZHVH04XB5YL1A/Slide1.jpg Impressions - 3. Clinical Application https://www.analyzemydata.org/impressions/2018/7/15/2-user-interface monthly 0.5 2018-08-08 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1533675067349-88KRRWQSS16WVIG5EZ2A/1*iaSvAm4v2OncAmGZVyOsRQ.png Impressions - 2. User Interface https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1533583810182-QSROW6IJ0GFE7G1GQT87/Model-View-Controller-High-Level-Diagram.png Impressions - 2. User Interface Figure 1. Illustration of the MVC model. I like this model because it shows how the Controller is the interface between the View and the Model, which interact with the user and the data/database, respectively (Note that other representations inaccurately show the Controller as directly interacting with the User, which is not really how MVC works. https://www.analyzemydata.org/impressions/2018/7/15/1-sensors-and-monitors monthly 0.5 2018-08-08 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1533674916687-0MQRMLEJPZR2AZHZNJWV/bigdataforbe.jpg Impressions - 1. Data Collection https://www.analyzemydata.org/impressions/2018/7/15/components-of-an-individualized-medicine-approach monthly 0.5 2018-08-08 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1531685656793-THBUGWSYPJMPAP9UR3Z6/board-chips-circuit-343457.jpg Impressions - Components of an Individualized Medicine Approach https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1531686656460-DO6RGXIV54ZSJ99RIR5X/animal-chihuahua-cute-39317.jpg Impressions - Components of an Individualized Medicine Approach https://www.analyzemydata.org/impressions/2018/4/27/impressions-overview monthly 0.5 2018-06-06 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1528316168278-YG6Y580Z5H6OHC57FPBL/pexels-photo-954623.jpeg Impressions - The Need for Individualized Approaches to Medicine https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1528321764492-F5UG2AOCJZ8QW5DOSSEF/Blog_Image_081717-768x393-2.jpg Impressions - The Need for Individualized Approaches to Medicine https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1528321319643-V49T56BW2E71M8X3WMJK/500369068.jpg Impressions - The Need for Individualized Approaches to Medicine https://www.analyzemydata.org/projects daily 0.75 2022-12-28 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1554925261352-E3MNT8OZMCK6U3J5S0MS/blur-close-up-conceptual-1194775.jpg Ongoing Projects https://www.analyzemydata.org/projects/2019/4/10/use-of-a-personalized-medicine-app-in-management-of-chronic-recurrent-conditions-a-pilot-research-study monthly 0.5 2022-12-28 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/e5c1922d-cc97-4043-a8a8-a320e8e945c8/iMTracker_1024.jpg Ongoing Projects - Use of a Personalized Medicine App in Management of Chronic Recurrent Conditions https://www.analyzemydata.org/projects/2019/4/10/machine-learning-to-improve-clinician-utilization-of-healthtech-innovations monthly 0.5 2022-12-28 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/998c6a72-72bd-409c-a73b-b15b00e7116c/physician-burnout-stock.jpg Ongoing Projects - Machine Learning to Guide Clinical Decision-making https://www.analyzemydata.org/analysis-types daily 0.75 2017-04-18 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492535881756-DEP4TETDJVA0Q80X8GFK/colorado-2064510_1920.jpg Analysis Types https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492535953269-34UIZMFHCDB04P11GI2C/landscape-1843128_1920.jpg Analysis Types https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492536401401-JONNJQX6WOS0SYZM00RW/autumn-1950876_1920.jpg Analysis Types https://www.analyzemydata.org/about-1 daily 0.75 2022-12-28 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1537281337754-EZM7N87IF0HBDADRUYTJ/udatalyze.0281c025-2.png About https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1537281662002-JHISB4JAJA1ELQOD9MN6/2000px-Fitbit_logo.svg.png About https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/f982686c-7e8a-474e-be31-b6dada41b578/Screen+Shot+2022-12-28+at+10.05.56+AM.png About https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/5a6d2818-ff46-4638-9e16-61a603482af8/Screen+Shot+2022-12-28+at+10.11.43+AM.png About - Make it stand out Whatever it is, the way you tell your story online can make all the difference. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492550966905-LJGVOPO2HFS1H59LY8NB/CUAnschutz_c_clr_rv.png About https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492539984231-1ZBA5GZYMVJ1NE9VNSSX/image-asset.png About https://www.analyzemydata.org/home daily 1.0 2022-12-28 https://images.squarespace-cdn.com/content/v1/573a04372fe131630fd25157/1469729752911-UFETSG39XPIFRSHOM0XH/hero_pattern.png Individualized Data Analysis Organization https://www.analyzemydata.org/criteria-for-analysis daily 0.75 2017-04-18 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492534534483-TTRV2DMJBOI449CCYDAS/image-asset.jpeg Criteria for Analysis https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492534949647-UO6KJJHO8OPHCGYEHCR1/image-asset.jpeg Criteria for Analysis https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492535461633-4VREVS2XTEYO3JM551AW/image-asset.jpeg Criteria for Analysis https://www.analyzemydata.org/backgroundreferences daily 0.75 2023-05-16 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492783738303-N0J8QKENT8FUL96EA0OB/image-asset.jpeg Background references This textbook was the first book I ever read on statistics. Excellent introduction to statistics and probability https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492784483885-G42L7SVMENAWTK759UWS/Kirkwood_book.jpg Background references This book was by far the best high-level explanation of both statistics and epidemiology. I still use references from it for teaching (particularly the explanation of confounding). https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492784518814-G1KPLFF1WDQ9MF2UMEJ1/Khan_academy.png Background references The lectures on linear algebra, calculus, and (of course) statistics, were an excellent supplement, and necessary to understand multivariate regression. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492784845060-MR6EGWTXO6K8EBOUAQOW/image-asset.jpeg Background references A challenging book as an introduction, with some advanced mathematical explanations, but nonetheless an excellent overview of regression approaches. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492785065869-1CD4XEMT029Q9P7VCJCS/image-asset.jpeg Background references A very practical explanatory approach to logistic regression by some of the pioneers of logistic regression (i.e., they have tests named after them). https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492785220659-4C6TMK69DUBOJXGFRSOD/image-asset.jpeg Background references As with the logistic regression textbook, this one also offers a very practical explanatory approach to survival analysis. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492785453539-03LHLE4GYU521Y2PORHI/image-asset.jpeg Background references This textbook provided a very comprehensive understanding of longitudinal analysis. The lecture slides from the website are a nice compliment, as was code from the UCLA statistics website, which is from a different textbook, but useful nonetheless. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492785824978-Z0D60S7T0FMZPT1SZOOF/image-asset.png Background references This textbook was one of the best I've come across for any topic, and presents Bayesian statistics in an understandable manner. The website also includes datasets and high-level code. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492804323719-T6L7B76VA2A45BZ5D064/image-asset.jpeg Background references This textbook is a very comprehensive, and somewhat technical, description of the various methods of time series analysis. Would also suggest this excellent online resource as a supplement. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492804753226-2K4DTFP6UAIZYQPP5SCU/image-asset.jpeg Background references Most of the Statistical code we'll be posting will be in the stats package R, which can be obtained here. R is open source, and very accessible for writing scripts and sharing code, which makes it ideal for an open/sharing platform. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492805027412-MHMBYOOE3QFF22S4PORW/image-asset.jpeg Background references Stata is also a very excellent, and user-friendly statistical package that I use for much of my 'bread-and-butter' statistics work. It has drop-down menus, wonderful supportive documentation, and is a great way to get started with statistical analysis for those who don't have experience with coding. This textbook was useful for me getting started, but there are others available through the Stata website. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492805497227-0I1A9OJXM07AQMK2YJMF/image-asset.jpeg Background references This textbook was a tremendous resource for someone who had never done anything computationally prior. You will be amazed how useful some of the tricks (especially the regular expressions stuff), even if you don't work in a computational biology area. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492805771697-6GOB3NNJNC7GCW1BYWHB/image-asset.jpeg Background references The MIT OpenCourseWare selection of online courses is a tremendous resource. I personally took the Introduction to Computer Science course, with the accompanying textbook, which was well-taught and easy to follow. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1492806010297-B7ORHAFQGYXUVTX0IYYA/image-asset.png Background references I used several textbooks from the Big Nerd Ranch, including programming in Swift and iOS design. An excellent resource, easy to follow, and you get to design some cool apps along the way. https://www.analyzemydata.org/machine-learning daily 0.75 2018-03-30 https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1522439295507-P6OMQBIZV2CQY1VYHUKH/1*5ZuLCsB1KXEPgHu-qJ8WxQ.png Machine Learning https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1522440872879-HT7E4KOONJ4B9LOU3UHC/ISL+Cover+2.jpg Machine Learning This very manageable textbook is a perfect starting point for anyone interested in learning about machine learning. In addition to a very nice overview of many methods of 'shallow' learning, it also includes R code to run the models yourself. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1522440890187-E961M726HBSRB45AHKIN/41aQrQaPseL._SX331_BO1%2C204%2C203%2C200_.jpg Machine Learning This somewhat larger and more technical textbook is the more advanced version of Introduction to Statistical Learning, by the same authors. It does not include code references, and remains at the 'shallow' machine learning level, but nonetheless provides a very nice comprehensive overview of these approaches. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1522441035369-V1KGHUDOQ10EHYGUIHA2/41QjQMexFwL._SX309_BO1%2C204%2C203%2C200_.jpg Machine Learning Although I'm not sure HMM's are technically considered 'machine learning', I would strongly recommend this book, which I've mentioned in prior posts, if you plan to delve into deep learning approaches with recurrent neural networks or reinforcement learning https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1522442158516-Q996C8ZKOLOAUJASFMIZ/51gtG44GHfL.jpg Machine Learning Before you dive into any online sources, blogs, or other textbooks, I'd highly recommend starting with this very basic, but understandable textbook. Not only does it provide an outstanding introduction to concepts like stochastic gradient descent, but it also teaches you to use to use Jupyter notebook and Python in a step-by-step manner. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1522442717368-8GK3AUHBTLOZABK1IRLU/Michael_Nielsen_Web_Small.jpg Machine Learning Neural Networks and Deep Learning. This online textbook is definitely one of the best places to start for understanding deep learning, and a free resource at that. https://images.squarespace-cdn.com/content/v1/58ef9b9a414fb5737104dbba/1522442406183-Y02VX47VHOVH8OLPOBIL/516KsohG9XL._SX382_BO1%2C204%2C203%2C200_.jpg Machine Learning Outstanding resource for using sklearn for 'shallow' learning approaches with code to run in Python. The description of deep learning is outstanding, although much of the code did not work for me and eventually I resorted to using Keras to run the models with Tensorflow. https://www.analyzemydata.org/publications daily 0.75 2023-05-08 https://www.analyzemydata.org/references-1 daily 0.75 2023-12-05